KR20240053333A - Symbol typing and selection of predicted candidates - Google Patents

Abstract

The present invention facilitates the input of non-character phrases and sentences by applying the word prediction input method, which is generally used as a character input method in English-speaking culture, not only to letters but also to symbols and spaces. Furthermore, in the case of Hangul, it provides a way to dramatically reduce the size of the database by configuring grammatical blank keys.

Description

Predicted word selection and symbol input method of sentence prediction system {Symbol typing and selection of predicted candidates}

Prediction input method

As cell phones began to be used, text calls began to become common, and with the change to smartphones, text calls became more common than voice calls. Furthermore, smartphones function as a multimedia communication tool rather than simply transmitting messages, resulting in an increase in the input of non-words such as web addresses and emails. Unlike general text input, such non-word input is inconvenient because letters and symbols must be input alternately. Nevertheless, when using a computer's full keyboard, it is easy to input letters and symbols, but on mobile devices that require input using only two fingers, it becomes much more difficult than when inputting only letters. In particular, it is not difficult to input letters and symbols using a keypad consisting of 12 keys based on the numbers 0 to 9 and * and #. To input only alphabetic characters, the English culture uses the word prediction input method (Disambiguity Input Method) using a 12-keypad. However, inputting phrases with a mixture of alphabets and symbols mentioned above is inconvenient as the letters and symbols must be entered separately. Ham is still there. Against this background, the present invention was developed for a keypad consisting of existing 12 keys and uses the word prediction input method (Disambiguation Input Method), which is widely used in English-speaking countries, to enable input of words and symbols, numbers, and even space characters. We would like to provide an input system.

In the case of the qwerty keyboard, in which all letters and symbols are assigned to each key, web addresses or passwords that use a mixture of letters and symbols can be easily entered, but in cases where the keyboard input space is limited, such as in a smartwatch, a wearable device, it is limited. Keyboards with a large number of keys must be considered. Moreover, as a text input device required for devices for recently emerging VR (virtual reality) and AR (augmented reality), the qwerty keyboard has limitations in usability, and new input devices are required. Unlike smartphones, these wearable devices are used for limited purposes because there is not enough space for text input. Headsets and glasses developed for AR provide a sufficient screen compared to smartwatches, but it is difficult to provide a keyboard for text input because it is difficult to input through touch (tactile) like a smartphone. In this regard, the present invention seeks to provide an input system with about 10 keys that overcomes the limitations of wearable devices and has input efficiency.

As cell phones were used, voice-centered communication gradually developed into message-centered communication, and as a result, the qwerty keyboard used in computers was transformed into a 12-keypad due to the limited space of cell phones, and an appropriate input system was developed. In particular, in English-speaking countries (including China and Japan), cell phone keypads have ambiguous keys with multiple letters assigned to each key. A word prediction input method (disambiguation input method) was devised that uses non-confirmed keys with multiple letters assigned to one key and presses these non-confirmed keys once to extract and input words composed of the characters assigned to these keys. and was widely used. The word prediction input method is a method that reduces the number of presses by allowing you to enter the desired word by pressing only once a non-confirmed key with multiple letters assigned to it. Taking the keyboard shown in Figure 1 as an example, in the case of the multi-tapping method, when you want to input the word 'boy', press the 'abc' key twice to input 'b'. Press the 'mno' key three times to input 'o', and finally press the 'wxyz' key three times to input 'y' to input 'boy'. On the other hand, the word prediction input method uses the designated keys 'b', 'o', and 'y' that make up 'boy', that is, 'abc', 'mno', and 'wxyz' unconfirmed keys, respectively, to input 'boy'. By pressing them once in order, you can enter words that can be created by combining the letters assigned to these keys (e.g. 'any', 'amy', 'boy', 'box', 'cow', 'bow', 'cox ', 'coy', etc.) are extracted from the dictionary database and displayed as predicted words (indicated in italics within the key) as shown in Figure 2. The input is made by selecting the word 'boy' to be input among these words. For the word prediction input method, as shown in the example above, a dictionary word database in which words including 'boy' are registered must be prepared and used.

In the present invention, the keyboard is composed of non-confirmed keys containing letters, symbols, numbers, and even space characters so that content including not only words contained in such a dictionary database but also phrases, sentences, and even clauses can be input. This configuration provides an input system with the same input efficiency as a qwerty keyboard, even with a keyboard consisting of about 10 keys.

The configuration of the present invention is to enable the predictive input method to go from inputting simple words to inputting entire sentences, including phrases containing symbols and even blank characters ('space'). For example, to input the name "Arnold J. Toynbee", the general prediction word input method involves predicting and inputting 'Arnold', 'J.', and 'Toynbee' in that order from a dictionary database, and then inputting the space character 'space'. Complete the name entry by entering between words. However, the configuration of the present invention is that when the key input of the 'ambiguous key sequence' shown in Equation (3) [Equation (3-1) to Equation (3-6)] below is made, as shown in FIG. 5, 'Arnold J. The name 'Toynbee' is displayed as a predictive phrase (in italics) in the keyboard area and can be selected/entered.

'abc'-'pqrs'-'mno'-'jkl'-'def'- ...(3-1)

'Sym' (ambiguous key representing symbols containing 'space' characters) - ...(3-2)

'jkl'- ...(3-3)

'Sym'- ...(3-4)

'Sym'- ...(3-5)

'tuv'-'mno'-'wxyz'-'mno'-'abc'-'def'-'def' ...(3-6)

That is, while the word prediction input method consists of three processes of predicting and then selecting 'Arnold', 'J.', and 'Toynbee' as words, in the present invention, it is reduced to one process to ensure convenience and speed of input. It will add up. Even if you enter only the ambiguous key sequence of Equation (4), which mixes the ambiguous key corresponding to the first letter of each word and the 'Sym' key representing the symbol (corresponding to L11 in Figure 1a), 'Arnold J. Toynbee' is entered in the dictionary. If it is registered in the database, it is extracted as a predictive phrase and the phrase 'Arnold J. Toynbee' can be entered.

'abc'-'Sym'-'jkl'-'Sym'-'Sym'-'tuv' ...(4)

In this context, the phrase prediction input method of the present invention, which uses the 'Sym' key as the character amibiguous key, has the effect of simplifying word prediction input. Furthermore, the input process to predict 'Arnold Toynbee' as a prediction phrase can be simplified as shown in equation (5).

'abc'-'Sym'-'tuv'-'mno'- ...(5)

That is, for 'Arnold', enter only the first letter 'abc', then enter the 'Sym' key corresponding to 'space', and enter the letters of 'Toynbee' corresponding to the last name in order, and 'Arnold Toynbee' will be predicted as a phrase. A method by which it can be extracted is provided. This will be explained in more detail in Examples 5 and 6.

The configuration of the present invention applies not only to the phrases discussed above but also to sentences as follows. In other words, when the key input corresponding to equation (6) is made to input 'Where are you going?', the first letter of the first word that makes up the sentence including 'Where are you going?' is 'w, x, y, z', the first letter of the second word is one of 'a, b, c', the first letter of the third word is one of 'w, x, y, z', and the first letter of the fourth word is one of 'w, x, y, z' It provides extreme convenience by listing sentences that are one of 'g, h, i' and selecting/entering 'Where are you going?' among these sentences.

'wxyz'-'Sym'-'abc'-'Sym'-'wxyz'-'Sym'-'ghi' ...(6)

This is the word prediction input method. When a key input corresponding to the ambiguous key sequence 'abc'-'mno'-'wxyz' is made to input 'boy', the predicted words including 'boy' are 'any', 'amy'. , 'cox', 'box', 'cow', 'bow', 'cmx', 'coy', 'coz', etc. are listed to be selected/entered, and further, to enter 'important', an ambiguous key sequence ' This is the same way that even if key input corresponding to ghi'-'mno'-'pqrs'-'mno'-'pqrs'-'tuv' is made, not only 'import' but also 'important' can be predicted and entered. .

The efficiency of the above configuration is noticeable when applied to a smartwatch, a wearable device representing mobile devices after smartphones, as shown in Figure 9a. In the case of smartwatches, they have a circular shape rather than a square. Accordingly, a keyboard of the shape shown in Figure 9b is also provided to enable the convenience and fast input speed of character input on a smartphone to be implemented on a wearable device.

In this context, inputting web addresses, which are essential for Internet searches, can also be easily performed. Web addresses such as 'www.fda.gov' may also be included. There is an ambiguous 'Sym' key with symbols including '.' designated, so to enter 'www.fda.gov', if the key input is performed in the following equation (1), as shown in Figure 1(c) Likewise, 'www.fda.gov' is displayed in the prediction list (displayed in italics on the keyboard) and is entered when selected. In other words, you can enter 'www.fda.gov' with the key input corresponding to equation (1) without having to change from text mode to symbol mode to enter the symbol '.'.

'wxyz'-'wxyz'-'wxyz'-'Sym'-'def'-'def'-'abc'-'Sym'-'ghi'-'mno'-'tuv' ...... (One)

In other words, you can enter 'www.fda.gov' with just 11 keystrokes without converting from character mode to symbol mode. And the keyboard shown in Figure 1 can be replaced with a keyboard consisting of 9 keys shown in Figure 9, so as a result, mixed phrases of letters and symbols such as web addresses and passwords can be easily entered with 9 keys, as in the qwerty keyboar. We would like to provide a way to do this.

In addition, the details of the present invention will be described through definitions of terms to be used below.

Terms example ambiguous key 'abc', 'def', 'ghi' ambiguous key sequence 'ghi'-'tuv'-'wxyz' numeric code '0', '1', '2' numeric code sequence '99902210368'

An ambiguous key is a key input of the predictive input method in which multiple characters are assigned to one key and when the key is pressed, it is set to mean all the specified characters instead of selecting one of the specified characters. That is, as shown in the second column of Table 0, all characters specified in the key are displayed within single quotation marks ('...'). The input sequence of these ambiguous keys is called 'ambiguous key sequence', and the conversion of ambiguous keys into numbers to convert them into numbers is called 'numeric code'. The relationship between the ambiguous key in Table 1 and the corresponding 'numeric code' is set based on the key arrangement of the keyboard shown in Figure 1a, and when the key arrangement of the keyboard changes, the relationship in Table 1 also changes. Therefore, the 'ambiguous key sequence' expressed as an ambiguous key is converted into a 'numeric code sequence' using 'numeric code'. Therefore, the 'ambiguous key sequence' in equation (1) is converted to the numeric code sequence in equation (2) below. '99902210368' ..... (2)

<Conversion relationship between ambiguous key and number code> ambiguous key
'abc'
'def'
'ghi'
'jkl'
'mno'
'pqrs'
'tuv'
'wxyz' numeric code
One
2
3
4
6
7
8
9 ambiguous key
'Num'
'Sym' numeric code
5
0

The predictive input system examined above has been used in English-speaking languages based on the alphabet. Therefore, when applying the predictive input system to Hangul, difficulties arise in using it. The reason is that, unlike alphabet languages, in Hangul, words used as subjects and verbs are combined with particles and endings, resulting in thousands of changes. This is because the amount of dictionary database for the predictive input system is enormous. For example, in the case of English, when looking at the sentence 'I went to school.', 'went', which is used as a verb, does not change as a verb. However, in the case of Hangul, when looking at the corresponding sentence 'I went to school', the verb 'went' is an ending that combines with the stem 'ga', such as '-was', '-was', '- In addition to 'Jiman', there may be hundreds or even thousands of them. In English, the meanings of these endings are conveyed by adding separate words. 'Have gone' is composed of 'have gone', 'had gone' is composed of 'had gone', and 'though' is composed of 'Although (I) went', so 'have' and 'had' are already registered in the dictionary database. ' and 'although' are used to indicate the meaning, so the number of words added to the dictionary database does not increase. On the other hand, in the case of Hangul, 'went', 'went', and 'gone' must be registered as separate words, which has the disadvantage of increasing the size of the database. There is no problem if there are only one or two, but if the number reaches thousands through suffix changes, the problem arises as the size of the database increases thousands of times.

In the case of Hangul, the same problem occurs with subjects. As can be seen in the sentence ‘Flowers blossom.’ in English, if the subject is ‘flower(s)’ and it is divided into singular and plural, no further words are added, whereas in Korean it is ‘Flowers blossom.’ ', 'flowers', 'flowers', 'flowers', 'flowers too', 'to flowers', 'like flowers', 'flowers', 'flowers', etc., nouns and particles are used as pronouns. A problem arises in that the number of words that must be combined and included in the dictionary database as one word increases in proportion to the type and number of searches. In other words, if the number of particles that can be combined with an adjective noun increases into the thousands, the amount of words that must be registered in the dictionary database increases thousands of times, causing the problem of a proportional increase in the capacity of the storage device for data storage and the actual A problem arises where the time it takes to search the dictionary database increases thousands of times. In the case of English, the particles that are combined with Korean nouns or the endings that are combined with verbs use separate words to convey their meaning. Therefore, words corresponding to Korean particles and endings are registered in a private database, so that they can be combined with other words differently from Hangul. Since they are not combined to create another compound word, the amount of the dictionary database is relatively small. For example, if the vocabulary 'Seoullo' is expressed in English, it becomes 'To Seoul'. In English, 'To' and 'Seoul' are separated by a space ('space'), so 'To' is one in the dictionary database. It is registered as a word and combined with words representing other places, it is used like 'To Busan', 'To Inchon', etc. In the word prediction input method, 'To', 'Busan', and 'Inchon' are each predicted as words, so 'To' There is no need to register 'Busan' and 'To Inchon' as separate vocabularies in the dictionary database. On the other hand, 'Seoullo' is not predicted as a word because 'Seoul' and 'Ro' are not distinguished from each other like in English, so to predict 'Seoullo', 'Seoullo' is used. must be registered in the dictionary database, and 'To Busan' and 'To Incheon' must be registered in the dictionary database respectively to be able to predict them as words. Therefore, rather than registering the words created by combining these nouns and particles in the dictionary database one by one, like English, the nouns and particles are registered in the dictionary database as different words, and a grammatical space is created so that they can be used like English. The present invention provides a method for reducing the number of Korean words included in a dictionary database.

Until now, in the information and communications field, the rules for input/output of characters on computers have defined each character as a code, and input/output has been performed using this code. However, in the present invention, a unique code ('numberic code' in Table 1) as shown in Table 1 is assigned to the non-confirmed keys 'abc', 'def', etc., where multiple characters are designated, and used for input/output of characters. Get the input system. However, what is different from before is that the non-confirmation key also includes the 'space' character, providing a way to easily output content expressed as letters, numbers, and symbols, especially for tasks such as Internet searches. This provides convenience of input so that you can easily find content that exists on the Internet with only about 10 keys. In other words, it provides an input system that can replace the qwerty keyboard.

And in the case of Hangul, the dictionary database for the predictive input system became large due to the particle suffix and the change in the ending of the verb, making its use difficult. However, by creating a 'grammatical gap', the 'postposition' of the noun and the 'suffix' of the verb are recorded as independent words in the dictionary database, making it possible to easily use the word prediction input system in Hangul.

Figure 1 - Key input sequence for entering 'www.fda.gov' and the state displayed on the screen as predicted by key input when 'www.fda.gov' is registered in the database
Figure 2 - A situation where predicted words are displayed on the screen after completing the key input to input 'boy'
Figures 3 and 4 - Process of entering '123@ivy.net' according to the configuration of the present invention
Figures 5, 6 - Process of entering 'Arnold J. Toynbee' according to the configuration of the present invention
Figure 7 - Process of entering 'Where are you going?' using the sentence prediction input method
Figure 8 - Process of entering 'www.fda.gov' into the search box using the phrase prediction input method
Figure 9 - A 9-key keyboard for sentence prediction input applied to a smartwatch.
Figure 10 - A non-expanded key input keyboard in the form of a keyboard with 9 keys, where the 'sym' [designated with a symbol and a 'space' key] key and the menu expansion key are assigned to the keys to which no letters are assigned.
Figure 11 - Example showing part of a Chicago area address book.
Figure 12 - Process of registering and entering “My name is HanAll.” into the database
Figure 13 - Symbol input process in a state where sentence prediction is made
Figure 14 - Configuration of a circular keyboard with independent 'space' and 'sym' keys
Figure 15 - Rectangular keyboard with independent 'space' and 'sym' keys
Figure 16 - Remote control with 8-way keys

<Input ‘boy’ using word prediction input method>

Figure 2 shows a prediction including 'boy' on the touch screen screen when keys are pressed in order according to the ambiguous key sequence ['abc'-'mno'-'wxyz'] to input 'boy' using the word prediction input method. It shows the state in which the word is displayed. In these predicted words, the first letter of each word is one of 'a, b, c', the second letter is one of 'm, n, o', and the third letter is one of 'w, x, y, z'. It is a word made up of one word, which is stored in a dictionary database and is extracted and displayed according to the frequency of use.

Figures 2(a) and (b) show a case where the predicted word is displayed inside the keypad, and Figure 2(c) shows a case where the predicted word is shown in a separate area above the keypad. In the case of Figure 2(c), to input a predicted word, touch the area where the word you want to enter is displayed among the list of predicted words displayed at the top of the keypad. In the case of Figures 2(a) and (b), the predicted word is displayed inside the keypad, so touching the area where the predicted word is displayed corresponds to the action of pressing each key, so in this case, the method for entering the predicted word If you move from the black key in the center of the keypad to the key with the predicted word displayed by dragging it and then lift your finger, the predicted word displayed on that key is entered. In the case of Figures 2(a) and 2(b), input can be made even if the key displayed with the predicted word is touched and the finger moves a certain distance regardless of the direction and then moves away. When a prediction word is input by one of the two methods above, the central black key (L22) can be used instead of the 'Sym' key (L21) in FIG. 2. The word prediction input method shown in FIGS. 2A and 2B has the advantage of reducing the keyboard input area because it does not require a separate area for displaying the predicted word compared to the method shown in FIG. 2C. An example of utilizing these advantages is the keyboard on a smartwatch shown in Figure 9.

Also include a web address such as 'www.fda.gov'. And among the keys used for word search, in addition to the keys designated with letters, there is an ambiguous 'Sym' key designated with symbols including '.', so to enter 'www.fda.gov', the keys are in the order of equation (1) below. When input is made, as shown in Figure 1(c), 'www.fda.gov' is displayed as a prediction list (displayed in italics on the keyboard) and can be entered when selected. In other words, you can enter 'www.fda.gov' with the key input corresponding to equation (1) without having to change from text mode to symbol mode to enter the symbol '.'.

'wxyz'-'wxyz'-'wxyz'-'Sym'-'def'-'def'-'abc'-'Sym'-'ghi'-'mno'-'tuv' ...... (One)

In other words, you can enter 'www.fda.gov' with just 11 keystrokes without converting from character mode to symbol mode. And the keyboard shown in Figure 1 can be replaced with a keyboard consisting of 9 keys shown in Figure 9, so as a result, mixed phrases of letters and symbols such as web addresses and passwords can be easily entered with 9 keys, as in the qwerty keyboar. We would like to provide a way to do this.

<Input of email address '123@ivy.net' using phrase prediction input method>

Figure 2(c) shows the keyboard configuration of a general word prediction input method. In order to enter the email address '123@ivy.net' in this general word prediction input method, the following process is performed.

(i) Enter ‘123’ in number mode

(ii) Switch to symbol mode and enter ‘@’

(iii) 'Switch to English mode and enter 'ivy' as word prediction input.

(iv) 'Switch to symbol mode and use '.' input

[On one of the keys located in the leftmost column of the keyboard in Figure 2(c)

If '.' is specified, symbol mode switching can be omitted.

* In this case, there is no need to switch to English mode in process (v)]

(v) Switch to English mode and enter ‘net’ as word prediction input.

Even if the input of '123@ivy.net' is done through word prediction input, the process requires a mode change, making input inconvenient. In contrast, if the phrase prediction input method is used with the keyboard configuration shown in Figure 1, '123@ivy.net' is input without changing the mode. The process is that when a key input follows the ambiguous key sequence corresponding to equation (7) below, the predicted phrase '123@ivy.net' is displayed in the keyboard area as shown in Figure 3(a), and '123@ivy. Select/enter 'net'. The method of selecting '123@ivy.net' is the same as the method of selecting the predicted word 'boy' in Figure 2. That is, move from the black key in the center of the keypad to the key marked '123@ivy.net' by dragging, then lift your finger or touch the key marked '123@ivy.net' and move a certain distance regardless of the direction. This is a method of entering '123@ivy.net' by moving your fingers apart.

'Num'-'Num'-'Num'-'Sym'-'ghi'-'tuv'-'wxyz'-'Sym'-'mno'-'def'-'tuv' ..... (7)

Here, the 'Num' key refers to L31 in Figure 3 and is an ambiguous key with 10 numbers assigned from 1 to 0.

<How to register a phrase in a dictionary database>

As shown in Example 1, the phrase prediction input method of the present invention provides a method for fast input without mode conversion ("character mode <-> symbol mode") compared to the multi-tapping method. Despite these advantages, one disadvantage of the predictive input method is when the word (phrase) to be input is not registered in the dictionary database. And the point is that you cannot know whether the word (or phrase) you want to input has been registered until the key input is actually completed. As a way to overcome this, we would like to explain how to complete the input by registering the phrase in the database even if the phrase you want to input is not registered in the database when the ambiguous key input is completed.

Figure 4 shows the process of registering '123@ivy.net' in the database when '123@ivy.net' is not registered in the database. In fact, the registration process in Figure 4 begins in the situation shown in Figure 3(b). In other words, the key input corresponding to equation (7) was entered to input '123@ivy.net', but no phrase, including '123@ivy.net', is registered in the dictionary database. So, as shown in Figure 3(b), instead of the predicted word, the numeric code sequence '55503890629' is displayed in the keyboard area, and the phrase 'Reg' is displayed together like a predicted word. The phrase 'Reg' is responsible for performing the registration process shown in Figure 4, and is always displayed on the keyboard even when the predicted word is extracted to prepare for the registration process. The reason is to prepare for cases where the phrase corresponding to the entered ambiguous key sequence has not already been registered in the dictionary database. And the method of selecting the 'Reg' phrase to execute the registration process is the same as the method of selecting the prediction word in Figures 2A and 2B. Therefore, when the phrase 'Reg' is selected, the registration process is executed as shown in Figure 4. If this registration process is explained in the case of '123@ivy.net', the corresponding extensions are activated in order and input is performed according to the numeric code that makes up the numeric code sequence '55503890629'. For reference, Table 2 shows the configuration of the extended keyboard corresponding to the numeric code.

<Contents of extended keyboard corresponding to numeric code> numeric code Supported extended keyboard content One Consists of ‘a’, ‘b’, ‘c’, ‘A’, ‘B’, and ‘C’ keys 2 Consists of ‘d’, ‘e’, ‘f’, ‘D’, ‘E’, and ‘F’ keys 3 Consists of ‘g’, ‘h’, ‘i’, ‘G’, ‘H’, and ‘I’ keys 4 Consists of 'j', 'k', 'l', 'J', 'K', and 'L' keys 5 Consists of keys ‘1’, ‘2’, ‘3’, ‘4’, ‘5’, ‘6’, ‘7’, ‘8’, ‘9’, and ‘0’ 6 Consists of ‘m’, ‘n’, ‘o’, ‘M’, ‘N’, and ‘O’ keys 7 Consists of 'p', 'q', 'r', 's', 'P', 'Q', 'R', 'S' keys 8 Consists of ‘t’, ‘u’, ‘v’, ‘T’, ‘U’, and ‘V’ keys 9 Consists of 'w', 'x', 'y', 'z', 'W', 'X', 'Y', 'Z' keys 0 32 symbols specified in qwerty keyboard and ‘space’ character (space)
` ~ ! @ # $ % ^ & * ( ) - _ = + [ ] { } ; : '" , . ? <> / | ＼

Against this background, the first step in the registration process of '123@ivy.net' is an extension version corresponding to the first numeric code '5' of the numeric code sequence '55503890629' of '123@ivy.net', as shown in Figure 4a. This is the process of activating the numeric keypad and entering the first letter '1' of '123@ivy.net'. [ In Figure 4, the dotted circle indicates a pressing operation. ] Subsequently, in Figures 4b and 4c, the second and third numeric codes of the numeric code sequence '55503890629' are '5', so the numeric keyboard is activated as in Figure 4a, and the second and third numeric codes of '123@ivy.net' are displayed. This is the process of entering the characters '2' and '3'. Figure 4d shows the process for inputting '@', in which the symbol keyboard corresponding to the fourth ambiguous key code '0' of the numeric code sequence '55503890629' is activated, and '$@~' including '@' on the symbol keyboard. 'It is the process of touching a key. The following is a registration process in the same context as Figures 4(a) to 4(d).

Figure 4e shows that when the '$@~' key is touched, '$', '@', and '~' are expanded left and right, and '@' is entered by removing the finger from the original key position. Figure 4f shows that in order to input 'i', the keyboard consisting of 'g', 'h', 'i', 'G', 'H', and 'I', which are the constituent alphabets of the 'ghi' key, is activated and 'i' The input of 'i' is completed by touching the designated key. When 'i' is entered, the keyboard consisting of 't', 'u', 'v', 'T', 'U', and 'V' is activated so that the next alphabet 'v' can be entered automatically. 'v' is input by touching the designated key. From Figure 4h to Figure 4m, the process of sequentially inputting 'y', '.', 'n', 'e', and 't'.

When this registration process is completed, as shown in Figure 4(m), the device is registered in the dictionary database internally and at the same time, '123@ivy.net' is entered into the input window (L41) and the user enters the phrase he/she wants to enter. There is no need to worry about whether or not you are registered in the database. After registration of '123@ivy.net', when a key input corresponding to the numeric code sequence '55503890629' (corresponding to equation (7)) is made, '123' is displayed in the keyboard area as shown in Figure 4(n). '@ivy.net' is displayed as a prediction phrase so that selection/input can be made and there is no need to register again.

<The process of entering a phrase containing a space character using the predictive input method>

The process of entering the name 'Arnold J. Toynbee' including a space character at once according to the configuration of the present invention is shown in Figure 5. As already explained in Example 2 of the input process of '123@ivy.net', the ambiguous key sequence for inputting 'Arnold J. Toynbee' using the keyboard shown in FIG. 1 is as shown in Equation (3). Here, the difference between 'Arnold J. Toynbee' and '123@ivy.net' is that space characters are treated like symbols, and ambiguous key input for space characters is done with the 'Sym' key.

When the key input process according to equation (3) is completed, 'Arnold J. Toynbee' is displayed as a prediction phrase in the keyboard area and is selected/entered as in Example 2. If there is no phrase corresponding to equation (3) in the dictionary database for the predictive input method, the numeric code sequence '17664204008696122' of equation (3) is displayed as a predicted word in the keyboard area, as shown in Figure 5b, and is stored in the dictionary database. It indicates that the phrase corresponding to the numeric code sequence '17664204008696122' is not registered. In this case, if you select the phrase 'Reg' shown in the keyboard area, the process of registering this phrase in the dictionary database is executed. Here, the method of selecting the 'Reg' phrase is the same as the prediction word selection method already described, and even if 'Arnold J. Toynbee' is already registered in the dictionary database, the 'Reg' phrase is displayed in the keyboard area so that it is always registered during the registration process. is to prepare to carry out. The reason is that the dictionary database registers only the most frequently used phrases, thereby reducing the database's proportion in the device's memory. Furthermore, by reducing the size of the database, the phrase prediction process can be processed quickly by registering all phrases in advance. This is preferable to eliminating the need for additional registration.

Figure 6 shows the process of registering 'Arnold J. Toynbee' in the dictionary database. In this registration process, the extensions corresponding to each numeric code that makes up the numeric code sequence '17664204008696122' of 'Arnold J. Toynbee' are activated in order according to the rules in Table 2, and the letters and symbols that make up 'Arnold J. Toynbee' are displayed. (including space characters) are entered in order. Specifically, in the state of FIG. 5A, if you move from the central black key to the 'Reg' key using a touch-drag operation and remove your finger from the keyboard, the first screen of the registration process shown in FIG. 6A will display 'a', 'b', and 'c. An extension containing ', 'A', 'B', and 'C' is displayed. When you touch the 'A' key, which is the first letter, 'A' is entered on the screen and the process of registering it in the database begins. When registration from Figures 6b to 6t is completed, 'Arnold J. Toynbee' is entered in the input window and simultaneously registered in the database. And after 'Arnold J. Toynbee' is registered in the database like this, when the key input is completed in the key sequence of equation (3), 'Arnold J. Toynbee' is displayed on the screen as a prediction phrase and can be selected, as shown in Figure 5a. /Enables input.

In order to enter the 'space' character included in the phrase 'Arnold J. Toynbee' during the process of registering the phrase 'Arnold J. Toynbee', the symbol keyboard shown in Figure 6g is activated and the black key in the center is touched. [During the registration process in Figure 6, the dotted circle indicates the position touched by the finger on the activated extended keyboard.] As shown in Figure 6h, a total of 9 symbols, including the 'space' key, are arranged and 'space' is displayed. Characters are assigned to the central black key, so they are entered by lifting your finger rather than moving it to the peripheral keys.

In this way, space characters are treated like letters or symbols, so phrases and even sentences containing space characters can be easily entered using the predictive input method.

<Sentence input using predictive input method>

As shown in Examples 1 and 3, numbers, symbols, and blank characters can be input using the predictive input method, and sentences can also be easily input using the predictive input method. For example, when entering 'Where are you going?' using a general word prediction input method, each word 'Where', 'are', 'you', and 'going' that makes up the sentence is extracted and entered into the dictionary database. While 'space' characters must be entered between each word, the predictive input method that treats the symbols of the present invention the same as letters is Equation (8) [Equation (8-1) ~ Equation (8-7) )], as shown in Figure 7(a), 'Where are you going?' is displayed in the keyboard area like a predicted word, and is selected and entered like a word.

'wxyz'-'ghi'-'def'-'pqrs'-'def' ... (8-1)

-'Sym' ... (8-2)

-abc'-'pqrs'-'def' ... (8-3)

-'Sym' ... (8-4)

-'wxyz'-'mno'-'tuv' ... (8-5)

-'Sym' ... (8-6)

-'ghi'-'mno'-'ghi'-'mno'-'ghi' ... (8-7)

-'Sym' ... (8-8)

< The process of easily registering a sentence in a dictionary database >

As shown in Figures 4 and 6, in the predictive input method, if the phrase (word) to be entered is not in the dictionary database, the method of registering it in the database is to activate the extensions corresponding to each number of the numeric code sequence in order. The letters, numbers, and symbols that make up the phrase were entered one by one in order. There is no problem with this method of registering a phrase if the phrase is short, but as in Example 4, if the sentence to be entered is not registered in the dictionary database and the sentence is long, it is quite inconvenient to register the characters that make up the sentence one by one. . A way to overcome this problem is to use a word prediction input method in the sentence registration process. That is, if the sentence “Where are you going?” is not registered in the sentence database after entering the ambiguous key corresponding to Equation 8, “Where are you going?” is not displayed as shown in Figure 7(b). Instead, the numeric code sequence '93272017209680363630' is displayed in the keyboard area like a predicted word. In this case, select the 'Reg' key shown in Figure 7(b) to enter the registration process of 'Where are you going?' The differences between the registration process shown in this example and the registration process in Example 2 are as follows. In Example 2, when the 'Reg' key is selected, the extended keyboard corresponding to the numeric code constituting the numeric code sequence is activated in order based on Table 2 to register the characters constituting the character one by one. On the other hand, in the registration process of this embodiment, if words or numbers constituting a sentence can be input through the word prediction method, only the parts excluding these words are registered one by one as in Example 2. Therefore, the first step in the registration process of this embodiment is the process of dividing the numeric code sequence of the sentence into symbol and non-symbol parts. In other words, in order to subdivide the numeric code sequence '93272017209680363630' into symbol and non-symbol groups, the ambiguous key code '0', which represents the symbol, is used as a boundary as follows.

[ '93272', '0', '172', '0', '968', '0', '36363', '0' ] ...(9)

For the parts other than '0' among these segmented numeric code sequences, a sentence registration process is performed using the word prediction input method. Equation (9) The first numeric code sequence '93272' of the segmented ambiguous numeric code sequence is a numeric sequence corresponding to a character, so as shown in Figure 7(c), the predicted word 'where' corresponding to the numeric code sequence '93272' is It is displayed, so if you select this as a method of selecting a predicted word, 'where' is registered as the first phrase of the sentence and entered into the input window at the same time. (L71)

[ In Figure 7(c), the first letter of 'Where', 'w', is displayed in capital letters. This means that 'Where' is changed to 'Where' by pressing the 'shift' key (L72) while 'where' is displayed on the keyboard as a prediction word. It is in a changed state. In Figure 7(c), the shift key (L73) is indicated in italics, which means that the 'shift' key (L73) has been pressed.]

If 'where' is not registered in the dictionary database, the number string '93272' is displayed in the area of the keyboard like a predicted word, as shown in Figure 7(d), and in this state, to execute the registration process of 'where', 'where' is registered. When the phrase 'Reg' is selected, the extension corresponding to the number string code '93272' is activated in order according to the configuration in Table 1, and the characters of 'where' are entered one by one in the same manner as shown in Examples 2 and 4. 'Where' is registered in the dictionary database and 'Where' is entered in the input window. (L74) When the step corresponding to '93272' is completed, the next step of 'Where' is automatically entered, the 'space' character (space). ) goes to the step of registering and the symbol keyboard is activated as shown in Figure 7(e). In this state, if you touch the black key in the middle and lift your finger, a blank character (space) is input, and the next step, the word prediction input step corresponding to the second number string '172' in equation (9), is executed, Figure 7 As shown in (f), 'are' is displayed as a predicted word in the keyboard area and can be selected/entered. Hereinafter, the registration process for 'you' and 'going' is carried out similarly, so that the sentence registration process is completed and 'Where are you going?' is entered in the input window at the same time.

[For reference, the process of registering 'Where' above in the dictionary database is performed as part of the sentence registration process, so the 'W' in 'Where' is registered as a capital letter, but 'Where' is not registered in the dictionary database. If not, it is desirable to register it in the dictionary database as 'where' in lowercase letters. This is because there is no problem if you later need to enter 'where' in lowercase letters, but if 'where' is registered as 'Where' as a word that is not part of a sentence, the word corresponding to the key input numeric code sequence '93272' Since 'Where' is displayed as a predicted word, in this case, there is no way to change the 'W' in 'Where' to lowercase, so a situation arises where 'where' must be distinguished from 'Where' and registered as a new word in the dictionary database. Because. ]

As seen above, when a word is registered as part of the sentence registration process, this has the effect of registering the word in the dictionary database at the same time as the sentence registration. Therefore, when applying the sentence registration process of this embodiment to the registration process of 'Arnold J. Toynbee' in Example 3, if 'Arnold' and 'Toynbee' are registered in the dictionary database, Figures 6(a) to 6(f) ) and the process of Figures 6(m) to 6(s) are reduced to one step each of predicting/selecting/entering 'Arnold' and 'Toynbee' in the dictionary database, so the dictionary registration and input process is It brings the advantage of simplification. If 'Arnold' and 'Toynbee' are not registered in the dictionary database, the process in Figure 6 can be continued.

< Simplified overview of phrase and sentence prediction key input method >

Examples 1, 3, and 4 illustrate a method of inputting not only symbols but also phrases and sentences including spaces using a predictive input method. We would like to describe a method to make key input simpler for the predictive input method of these phrases and sentences.

The ambigous key sequence for inputting 'Who is he?' as a sentence prediction input is equation (10-2) according to the method of Example 3. And, unlike equation (10-2), equation (10-3) corresponds to the first letters 'w', 'i', and 'h' of the words that make up 'Who is he?' in (10-1). This is an ambiguous key sequence in which only the ambiguous key is keyed in, and the sentence is predicted even if only the ambiguous key corresponding to the first letter of the word is entered. This means that even if an ambiguous key corresponding to 'import' is entered to predict 'important' in the word prediction input method, not only 'important' but also 'importance' and 'important' that match the first part are predicted in the dictionary database. It is in the same context as the method. The key to simple key input like Equation (10-3) is to exclude symbols and space characters from the elements that make up the sentence you want to input, and to ensure that the sentence is predicted even if only part of the remaining parts are entered.

Who is he? ...(10-1)

'wxyz'-'ghi'-mno'-'Sym'-'ghi'-'pqrs'-Sym-'ghi'-'def'-'Sym' ...(10-2)

'wxyz' -'Sym'-'ghi' -'Sym'-'ghi'- 'Sym'...(10-3)

Therefore, the criteria for extracting predicted sentences that allow sentences to be predicted even when only part of the words forming the sentence are keyed in as shown above are specified in Table 5, and these criteria are applied to the sentences in Table 3 and the ambiguous key input in Table 4 and registered in the database. We would like to explain the process by which the sentences in Table 3 are extracted as predicted sentences according to the ambiguous key input in Table 4.

<6-I. Calculate predicted sentence extraction variable numeric code sequence from sentences registered in the database>

As shown in Table 3, sentences registered in the database are expressed by the corresponding ambiguous key sequence, and the ambiguous key sequence expressed in this way is subdivided with the 'Sym' key as a boundary. These subdivided segments are called segments and these segments are shown in Table 3. Arrange them in order starting from the top of the first row. Then, a numeric code sequence is calculated from the ambiguous key sequence of each segment according to Table 1, and this numeric code sequence is entered in the corresponding segment in the second column of Table 3.

<6-II. Calculate predicted sentence extraction variable numeric code sequence from key input>

As in paragraph <6-I> above, a numeric code sequence is calculated based on Table 1 from the ambiguous key sequence of the key input for sentence prediction and then entered into the corresponding segment.

The numeric code sequence calculated in paragraph <6-I> (corresponding to the numeric code sequence of the sentence in the database) and the numeric code sequence calculated in paragraph <6-II> (corresponding to the numeric code sequence of the key input) are shown in Table 6. Comparison is made according to the criteria, and if both criteria 1 and 2 are satisfied, it is determined that the key input for sentence prediction matches the sentence in the database, and the sentence in the database is extracted as a predicted sentence. By comparing the numeric code sequence of 'Where are you going?', which is a sentence registered in the database in Table 3, with the numeric code sequence of the ambiguous key input in Table 4, the key input in Table 4 is the sentence in the database in Table 3. We would like to explain the process of determining whether or not can be extracted using the standards in Table 5.

In other words, when comparing the numeric code sequence of key sequence 1 (hereinafter referred to as 'A') in Table 4 and the sentence in Table 3 (sentence registered in the database - hereinafter referred to as 'B') according to the standards in Table 5, the results are as follows. same.

When comparing the number of digits of the numeric code sequence of A and B for each segment based on standard 1, in the case of the first segment, the numeric code sequence of B is '93272', so the number of digits is 5, and the numeric code sequence of A is '932', so the number of digits is 5. becomes 3, which is less than the number of digits in B's numeric code sequence, so it satisfies the standard. Hereinafter, from the 2nd to the 7th segment, the number of digits in A's numeric code sequence is less than or equal to the number of digits in B's numeric code sequence, thus satisfying the standard. Lastly, in the 8th segment, since there is no numeric code sequence of key input, the number of digits is 0, and since the numeric code sequence of the database sentence is '0', the number of digits is 1, thus satisfying the criterion of 1.

Based on criterion 2, when comparing whether the numeric code sequences of A and B for each segment match from the front, in the case of the first segment, the numeric code sequence of B is '93272' and the numeric code sequence of A is '932', so they match from the front. Therefore, it satisfies the standard. Subsequently, when comparing from the second segment to the seventh segment, the numeric code sequence of A matches the numeric code sequence of B from the front, so the standard is satisfied. And in the case of the 8th segment, since there is no number code sequence for key input, the criterion is satisfied, so criterion 2 is also met. Therefore, key sequence 1, corresponding to the key input in Table 4, satisfies the criteria in Table 5, allowing 'Where are you going?' to be extracted from the database as a predicted sentence and selected/entered.

Just as key sequence 1 in Table 4 was compared with the sentence in the database above, when comparing key sequence 2 in Table 4 with the sentence 'Where are you going?' in Table 3, the sentence in Table 3 only has up to the 8th segment, whereas key sequence 2 Since there is a 9th segment, the condition of criterion 2 is not met, so 'Where are you going?' cannot be extracted as a prediction sentence.

Where are you going? ambiguous key sequence number code sequence Segment 'wxyz'-'ghi'-'def'-'pqrs'-'def'- 93272 1st 'Sym'- 0 2nd 'abc'-'pqrs'-'def' 172 3rd 'Sym' 0 4th 'wxyz'-'mno'-'tuv' 968 5th 'Sym' 0 6th 'ghi'-'mno'-'ghi'-'mno'-'ghi' 36363 7th 'Sym' 0 8th

key sequence 1 key sequence 2 ambiguous key sequence number code sequence Segment ambiguous key sequence number code sequence 'wxyz'-'ghi'-'def' 932 1st 'wxyz'- 9 'Sym'- 0 2nd Sym'- 0 'abc'-'pqrs'-'def' 172 3rd 'abc'- One 'Sym' 0 4th Sym' 0 'wxyz'-'mno'-'tuv' 968 5th 'wxyz'-'mno'-'tuv' 968 'Sym' 0 6th Sym'- 0 'ghi'-'mno'-'ghi' 363 7th 'ghi'-' 3 8th 'Sym'- 0 9th 'abc'- One

Standard 1
When comparing the number of digits in the number code sequence in each segment
The number of digits in the number code sequence of key input is
The number of digits in the number code sequence of the database sentence must be less than or equal to.
(However, if there is no number code sequence in the segment, the number of digits is treated as 0 (zero) and compared.)

Standard 2
When criterion 1 is met,
In each segment
The number code sequence of key input is
It must match the number code sequence of the database sentence by comparing it from the front.
(However, if there is no number code sequence for key input, criterion 2 is met.)

key sequence 3 key sequence 4 ambiguous key sequence number code sequence Segment ambiguous key sequence number code sequence 'wxyz'- 932 1st 'wxyz'- 9 'Sym'- 0 2nd Sym'- 0 'abc'-' 172 3rd 'abc'- One 'Sym'- 0 4th Sym' 0 'wxyz'-' 968 5th 'wxyz'-'mno'-'tuv' 968 'Sym'- 0 6th 'ghi'- 3 7th

By applying the standards in Table 5 as discussed above, key sequence 3 and key sequence 4 shown in Table 6 satisfy the standards in Table 5 when compared to the sentences in Table 3 (assumed to be sentences in the database), so key sequence 3 and Key sequence 4 extracts 'Where are you going?' as a predicted sentence from the database and allows it to be selected/entered. In other words, by entering only the first letter of each word in 'Where are you going?', it can be extracted as a predicted sentence (in the case of key sequence 3), even when an ambiguous key corresponding to 'you' is entered in the middle of the sentence. (In case of key sequence 4) ‘Where are you going?’ is guessed as a predictive sentence and selected/entered. In order to implement simple key input for this convenience, applying the standards in Table 5 to the process of extracting predictive sentences from the database makes it possible to extract predictive sentences even by inputting only part of the sentence, thereby implementing simple key input. .

<Enter simple web address>

The sentence prediction input method described in Example 6 can also be applied to web address input. For example, to enter the web address phrase "www.fda.gov", a key input generally represented by equation (11) must be made.

'wxyz'-'wxyz'-'wxyz'-'Sym'-

'def'-'def'-'abc'-'Sym'-'ghi'-'mno'-'tuv' ....(11)

However, if the simple input method of the sentence prediction input method of Example 6 is applied, 'www.fda.gov' can be extracted from the dictionary database and selected/entered even by key input corresponding to equation (12).

'wxyz'-'Sym'-'def'-'def'-'abc'-'Sym'-'ghi'-'mno'-'tuv' ....(12)

Web addresses generally start with 'www.', so even when keyed in as in equation (12), 'www.fda.gov' can be extracted and entered from the dictionary database.

< Internet search and dictionary database utilization using predictive input method >

Until Example 7, the phrase or sentence prediction input method was input through the process of extracting sentences from a dictionary database. And the word prediction input method has been used as an input method for mobile devices with a limited number of keys. Since most of these mobile devices are connected to the Internet, using a dictionary database stored on an Internet server for the word prediction input method is more efficient than using a dictionary database built into the mobile device. In the case of words, a dictionary database can be built with several megabytes of memory, but in the case of sentences, much larger memory capacity is required, so it is advantageous to use the memory of an Internet server rather than storing the database on a mobile device. In the query process of extracting data from a database, using the high-performance processor of an Internet server allows the input process to be processed more quickly than performing the query process on a mobile device. This means that the Internet transmission speed of mobile devices reaches gigabytes per second, and the process of transmitting processing results from the Internet server to the mobile device becomes close to real-time, so there is no need to have a database on the mobile device.

If the sentence prediction input method can be applied to searches on Internet search engines such as google, yahoo, and bing, it will provide an easy way to search even in cases where the number of keys is limited and input is difficult using general input methods, such as smartwatches. It is done. However, for this purpose, the index of the database being built by each search engine must be constructed as shown in Table 7. In other words, the index corresponding to a word (phrase, web address, sentence, etc.) must be constructed not only with a general string code but also with an ambiguous key sequence. The numeric code sequence in Table 7 is calculated based on Table 1, so if the contents of Table 1 change, it must be changed accordingly.

string code ambiguous key sequence numeric code sequence
according to
Table 1 "www.fda.gov" 'wxyz'-'wxyz'-'wxyz'-'Sym'-'def'-'def'-'abc'-'Sym'-'ghi'-'mno'-'tuv' '99902210368' "www.efa.gov.in" 'wxyz'-'wxyz'-'wxyz'-'Sym'-'def'-'def'-'abc'-'Sym'-'ghi','mno','tuv'-'Sym'-'ghi ','mno' '99902210368036' “We will be back.” 'wxyz'-'def'-'Sym'-'wxyz'-'ghi'-'jkl'-'jkl'-'Sym'-'abc'-'def'-'Sym'-'abc'-'abc '-'abc'-'jkl'] '920934401201114'

If the database of the search engine is built as shown in Table 7, an ambiguous key corresponding to equation (13) is entered in the input window of the search program and executed, and a link to the U.S. Food and Drug Administration homepage site is created as shown in Figure 8 ( link) You can see a list of items and select the address ('www.fda.gov') displayed in the keyboard area, or touch and select 'www.fda.gov' from the list sites displayed on the screen. .

'wxyz'-'wxyz'-'wxyz'-'sym'-'def'-'def'-'abc'-'sym'-'ghi'-'mno'-'tuv'

...(13)

In this way, even with the keyboard shown in Figure 1, you can easily connect to Internet sites and even enter sentences easily. Smartwatches, which are currently attracting attention as a next-generation mobile device for smartphones, can be used to search the Internet, enter commonly used sentences, and even enter passwords just as easily as on smartphones.

<Application of non-confirmed key keyboards to touch-type devices and push-type devices>

The sentence prediction input method examined in the above embodiments was examined for screen touch key input, but can also be implemented when applied to a push keypad. The only difference is that in the case of the touch method, the drag action is given the function of selecting a predicted word, so touching the key is ambiguous key input, and the drag action is responsible for the function of selecting/inputting the predicted word or predicted sentence. However, since a drag operation is not possible in the keypad push input method, the sentence prediction input method of the present invention can be applied to a keypad with a push key by replacing the drag action with a long press operation instead.

For the sentence prediction input method of the present invention, a symbol or a space character is input by designating it as an ambiguous key like a character. However, in the case of space characters, instead of having a separate ambiquous key, the number of keys can be reduced by touching and dragging the key to input the character specified for the key and the space character at the same time. In other words, if you touch the 'abc' ambiguous key and move it over a certain distance, it performs the same function as entering the 'abc'-'Sym' key. In this case, the function of selecting/entering a predicted word (sentence) by touching the 'abc' key and then dragging must be implemented in a different way.

<Display priority of search contents in non-confirmed search using non-confirmed key>

As seen in Example 8, the use of the input system using the non-confirmed key of the present invention is similar to Internet search, where there is no database of words, sentences, and phrases for ambiguous input on the user's device, and instead, a search engine for searching. Even if a database exists, it can be used as an input system. However, as seen in Example 8, in the case of Internet search, the amount of expected search content cannot be predicted, unlike the limited content registered in the user database. Therefore, in order to increase search efficiency, the search content is extracted for non-confirmed key input. It is desirable to extract and display them in the following order. That is, in Example 6, the string code extracted for the ambiguous key sequence according to the non-confirmed key input is extracted according to the following priority order.

Priority 1:

If the number code sequence corresponding to the searched content exactly matches the entered number code sequence, it is extracted preferentially.

Priority 2:

The number code sequence corresponding to the searched content does not exactly match the entered number code sequence.

If it meets the criteria in Table 5, it is extracted preferentially.

Referring to the following Table 7 and looking at the extraction according to the above priority, when a non-confirmed key is input according to key sequence 3, the extracted sentences are all extracted sentences 1 to 3 according to the standards in Table 5, but priority 1 is selected among them. Accordingly, extracted sentence 3 is displayed as the highest priority, and extracted sentences 2 and 3 are displayed in order so that the user can select them. If only one extracted sentence is displayed, only extracted sentence 3 is displayed. And when the non-confirmed key is input according to key sequence 2, extracted sentences 2 and 3 are extracted according to the standards in Table 5, and among them, extracted sentence 2 is displayed preferentially. Therefore, key sequence 3 can extract the entire sentence 'My name is United.' even if you enter the non-confirmed key 'tuv', which corresponds to the 'U' of 'United', the last word of the sentence. However, among the extracted contents, extracted sentence 2 and When 3 is searched, there is a lot of search content extracted by entering the remaining letters of 'United' sequentially, so the process of selecting among them can be omitted.

Extracted sentence 1
My name is United . My name is United. Extracted sentence 2
My name is Uni . My name is Uni. Extracted sentence 3
My name is U . My name is U. key sequence 1 'mno'-
'wxyz' 'Sym' 'mno'-
'abc'- 'Sym' 'ghi'-
'pqrs' 'Sym' 'tuv'-
'mno'-
'ghi'-
'tuv'-
'def'-
'def' 'Sym' key sequence 2 'mno'-
'wxyz' 'Sym' 'mno'-
'abc'- 'Sym' 'ghi'-
'pqrs' 'Sym' ''tuv'-
'mno'-
'ghi'- 'Sym' key sequence 3 'mno'-
'wxyz' 'Sym' 'mno'-
'abc'- 'Sym' 'ghi'-
'pqrs' 'Sym' 'tuv' 'Sym'

< Configuration and advantages of a 9-key keyboard with a 'sym' key and a menu (expansion) key designated with symbols and 'space'>

Figure 10 shows a circular keyboard and keyboard arrangement for entering an unconfirmed search. The characteristic of this keyboard is that it has separate menu keys and 'Sym' keys, and the arrangement of the character keys maintains the arrangement of the qwerty keyboard. The character keys at the top are divided into three, 'qwer', 'tyu', and 'iop'. It has three non-confirmed keys. The middle and bottom are composed of two keys each, dividing the consonant arrangement of the qwerty keyboard into left and right. Moreover, numbers are also assigned to character keys: '0' and '5' are assigned to the 'qwer' key, and '1' and '6' are assigned to the 'tyu' key. In other words, the number shown on the key and the number plus '5' are assigned to each key, making it possible to search for numbers without a separate key for numbers. And the 'Sym' key contains symbols corresponding to the numeric code '0' in Table 2, including 'space'. And the 'menu' key on this keyboard performs various functions. When one of them is touched, the 'category(,)' key is activated as shown in Figure 10(b), making it easier to extract phrases or sentences. It performs the function of doing so.

The following explains the process of entering a non-confirmed key using the 'category' key to search a database of US postal addresses. Figure 11 shows a portion of the address book in Chicago, USA. To make address notation clear, 'space' is indicated as '-' (hyphen) in FIG. 11. In the address shown in Table 11, '-' (hyphen) is converted to 'space' and displayed as actually used, as follows.

1723 WEST 107TH STREET,

City of Chicago, IL

60643

To search the above address, the addresses are arranged in a row as follows. The separated addresses indicated by ',' (comma) are 'zone separator' symbols to facilitate address distinction.

1723 WEST 107TH STREET, City of Chicago, IL 60643

Therefore, if you input this 'zone separator' symbol ',' (comma) not as a simple symbol but as the separator 'category' key in Figure 11, the ambiguous key input to search this address is Equation (14) This can be done together, but the unique thing is that using the 'category' (,) key brings the advantage of making search easier by abbreviating ambiguous key inputs more than simple inputs according to the standards in Table 5. In other words, even if the non-confirmed key corresponding to 'STREET' is not entered by entering the category key, it is notified that the non-confirmed key entered after the category key corresponds to the first letter of 'City of Chicago' indicating the city, corresponding to 'STREET'. Even if the non-confirmed key input is omitted, address search is possible. However, in order to use the category key, there is a limitation that regulations must be set in advance by which search contents must be searched according to these rules. However, if these rules are set, the category key can be used as shown in the keyboard configuration shown in Figure 11. The following Example 13 explains a search in which the category key of the keyboard configuration shown in FIG. 11 is actually utilized.

'1tuy'-'2iop'-'2iop'-'3asdf'- : '1723'

'sym'- :

'0qwer'- : 'W'(est)

'sym'-

'1tyu'-'0qwer'-'2iop'-'1tyu'-'ghjkl'- : '107TH'

'category'-

'4zxcv'- : 'C'(ity)

'sym'- :

'2iop'- : 'o'(f)

'sym'-

'4zxcv' : 'C' (hicago) ... (14)

The advantage of the 'category' key examined in Example 12 is that the sentence or phrase to be searched does not have to have a continuous structure. And in general, 'space', which connects two words in Internet searches, unlike 'space' in a sentence, does not mean that the two words are adjacent within the sentence, but rather means that the sentence or phrase contains the two words at the same time. We would like to distinguish the difference between the two meanings into 'sentence-space' and 'search-space'. In other words, the 'sentence-space' key extracts phrases or sentences in which the two words are adjacent when searching when two words are separated by 'sentence-space', and the two words separated by 'search-space' are used to extract phrases or sentences where the two words are adjacent to each other. In the case of words, sentences and phrases containing two words are extracted.

The advantage of dividing the 'space' key into a 'sentence-space' key and a 'search-space' key is that it allows you to accurately search for the sentence or phrase you want to extract. For example, in music search, if you want to search for the music title "Nocturnes, Op. 9 (Chopin)", rather than entering all key sequences corresponding to this title, enter the key sequences corresponding to 'Nocturnes' and 'Chopin'. Just enter it by combining it with ‘search-space’. However, in order to accurately search the phrase "chopin nocturnes", the key sequences corresponding to 'Chopin' and 'Nocturnes' must be connected in 'sentence-space' to exclude "Nocturnes, Op. 9 (Chopin)". Since only the phrase "chopin nocturnes" is extracted, the number of prediction sentences is reduced, making selection/input of prediction sentences easier.

According to the rules of Example 11, if there is no content to be searched, the search is not performed. In this case, if the 'category' key is used, there is no need to search for the non-confirmed key input key sequence in order, but the key sequence located before and after the 'category' key. are searched and extracted according to the standards in Table 5.

'b n m'-'i o p'-'z c xv'-' t yu'-'ty u' -' qwe r '-'b n m'-'qwe r '-'qw e r'- 'a s df' ..(15-a)

'category'-

'zx c v'-'g h jkl'-'i o p'-'io p '-' i op'-'b n m' ...(15-b)

Therefore, the key sequence of Equation 15 is the predicted words "Nocturnes" and "Chopin" corresponding to the key sequences of Equation 15(a) and Equation 15(b), separated by the 'category' key ('search-space' key). It extracts searched phrases and allows the user to select them. As an example, "Nocturnes, Op. 9 (Chopin)" satisfies the key sequence corresponding to 'Nocturnes' corresponding to equation (15-a) divided by the category key and to 'Chopin' corresponding to equation (15-b). It is searched by satisfying the corresponding key sequence. According to the configuration of this embodiment, when the 'space' key and the 'and-space' key are distinguished, if the user wants to search only the phrase "Nocturnes Chopin", if the 'space' key is entered instead of the 'category' key in Equation 15, " Phrases such as "Nocturnes, Op. 9 (Chopin)" are excluded and only the phrase "Nocturnes Chopin" is extracted so that the user can easily select or enter it.

The database used to input characters, phrases, and sentences through non-confirmed key input may have a dual structure of a user database and a server database. In the case of search, the database of the search engine is used, so it is necessary to use the user's database. There is no However, when an unconfirmed key is entered to input a character, phrase, or sentence on the user's device, the priority is set to first search the user database embedded in the user's device to detect the character, phrase, or sentence to be entered. This is the configuration of this embodiment.

<The difference between a keyboard configuration that has separate ‘sym’ and ‘menu’ keys and a keyboard configuration that uses the ‘sym’ and ‘menu’ keys simultaneously>

The keyboard type shown in Figure 10(a) is composed of a 'menu' (expansion) key and a 'sym' key for symbol input, and can be applied to both touch and key press type devices, while in Figure 9 The configuration of the keyboard shown is that the black key in the center must simultaneously perform the 'sym' key for symbol input and the 'menu' key to perform various functions, so the ambiguous key of the present invention is used on touch input devices. It is possible to perform a predictive input method using a key press type device, but in order to perform the functions of the 'sym' key and 'menu' key on the central black key, a simple pressing operation such as increasing the number of presses or pressing it for a long time is required. This entails the inconvenience of having to provide a method to distinguish it from the other. For example, in the case of the keyboard shown in Figure 9, in a touch type device, the black key in the center is simply touched. If a finger is raised, it is used as the 'sym' key function, and in the state where the finger is touching, the 'sym' key is used. ', the 'category' key is activated so that when you drag your finger to the 'category' key and then lift your finger from the keyboard, the 'menu' key performs the function of the 'menu' key, which performs the function of the 'category' key. It is possible to distinguish the functions of the 'sym' key. However, unlike touch-type devices, touch/drag operations are not possible in keypad input devices that use simple push keys, so the 'sym' key function is performed with a simple press action, and if the key press action continues for a long time, it performs the function of the 'menu' key. However, it is not easy to distinguish between an actual long press and a simple press, causing inconvenience. As shown in Figure 10(a), even if the number of character keys is reduced by one, the 'menu' key is replaced with the 'sym' key. In distinction from this, there is an advantage that the same keyboard operation configuration can be used for touch-type input devices and push-type keypad input devices.

This is the difference between the keyboard shape shown in Figure 10 and the keyboard shape shown in Figure 9. Figure 10(c) shows the keyboard of Figure 10(a) applied to a square device. The difference from the keyboard of Figure 10(a) is that the symbol and space character ('space') are designated as ambiguous 'sym'. The key is placed downward.

< How to register sentences in the user database using a 9-key keyboard with non-confirmed keys ‘sym’ and ‘menu’ keys>

This embodiment will explain the process of registering the sentence shown in Example 5 in the user database using the keyboard shown in FIG. 10(a). In Example 5, the keyboard consists of more than 10 keys, so it is easy to designate various function keys, including the 'shift' key, in addition to the letter keys. However, the 9 keys shown in FIG. 10 are used to perform these various function keys. The ‘Menu’ (expansion) key is used. The keyboard of FIG. 10 is configured to allow easy character input like a qwerty keyboard using only the 9 key using the menu (expansion) key. As an example, this configuration will be explained through the process of registering the sentence “My name is HanAll.”

'bn m '-'t y u'- (16a)

'sym'- (16b)

'b n m'-' a sdf'-'bn m '-'qw e r'- (16c)

'sym'- (16d)

' i op'-'a s df'- (16e)

'sym'- (16f)

'g h jkl'-' a sdf'-'b n m'-' a sdf'-'ghjk l '-'ghjk l'- (16g)

'sym' (16h)

Figure 12a shows that in order to input the sentence “My name is HanAll.”, the non-confirmed key key sequence of equation (16) corresponding to this sentence is input. However, because there are no predicted phrases (sentences) detected from the database, the predicted words are not displayed on the keyboard as shown in Figure 12(a), and instead, 'REG', which means 'sentence registration' possible status, is displayed in red to the right of the 'menu' key. It is displayed in colored letters. (The reason 'REG' is displayed in red means that you need to press the 'menu' key to activate the 'REG' key.) The reason there is no predicted sentence is that 'HanAll' is unique among the words that make up the sentence. This is because it was omitted as a noun, and if the proper noun 'HanAll' had been registered in the database, the sentence would have been extracted and displayed. Therefore, in order to register "My name is HanAll.", which includes 'HanAll', as a sentence in the database, the process of registering 'HanAll' in the dictionary database is also included. And the first process of registering "My name is HanAll." is to press the 'menu' key in the keyboard state shown in Figure 12(a) and activate the 'Registration' (REG) key as shown in Figure 12(b). Next, you need to press the 'REG' key.

When the 'Sentence Registration' (REG) key is activated and pressed, the sentence registration process begins. The first process is 'my', the first word of the sentence corresponding to the key sequence in equation (16a), as shown in Figure 12(c). ' is extracted and appears in the dictionary database. To make a capital letter, press the 'Cap' key. Then, as shown in Figure 12(d), the extracted words are changed to uppercase letters, and when you press the key designated as 'My' among the extracted words, 'My' is input. Next, the symbol and space keyboard corresponding to the 'sym' key corresponding to equation (16b) appear as shown in Figure 12(e). 'Space' is entered by pressing the key designated with the 'space' character. Next, when the extracted words corresponding to the key sequence corresponding to Equation 16(c) are displayed on the keyboard as shown in Figure 12(f), 'name' is entered by pressing the key area where 'name' is designated. After completing the input of the 'space' ('space'), 'is', and additional 'space' ('space') in the same manner as shown in Figure 12(f) ~ (h), 'HanAll' must be entered. However, unlike 'my', 'name', 'is', etc., 'HanAll' is not registered in the dictionary, so it is not displayed as an extracted word on the keyboard, and 'reg', which means that 'HanAll' is not registered and must be registered, is also displayed. It is shown in 12(j). Here, the difference between FIG. 12(j) and FIG. 12(b) is that the mark indicating registration is indicated in lowercase letters as 'reg' rather than 'REG'. That is, 'REG' in Figure 12(b) means that the sentence (or phrase) registration process must proceed because the sentence is not registered in the database, and 'reg' in Figure 12(j) means that registration of the word must begin. it means. Therefore, if you press the key designated 'reg' in Figure 12(j), the word registration process corresponding to Figures 12(k) to (p) proceeds. When the word registration process performed during the sentence registration process is completed, the input of '.' (period-period) corresponding to Equation 16(h) is finally made as shown in Figure 12(q), resulting in 'My name is HanAll.' input is made. When 'My name is HanAll.' is registered in the database in this way, when the key sequence corresponding to equation (16) is entered, it is displayed on the keyboard as an extracted sentence and can be selected/entered.

In the case of word prediction input using a non-confirmed key, prediction is possible in most cases because commonly used words are registered in a dictionary database. However, in the case of sentences and phrases, unlike words, all combinations of words connected with the 'space' key are prepared to create a database. Since registration is not possible, a method of direct registration was explained in Example 16 as a supplementary method. Nevertheless, in order to compensate for this, in this embodiment, a new sentence prediction method is constructed to easily supplement sentence registration. For example, if you want to input "My name is HanAll" in Example 16 using the sentence prediction method, when the key corresponding to Equation 16 is entered, the entire sentence cannot be searched in the database, so the registration process using the method in Example 16 is performed. In this configuration, before proceeding with the sentence registration process, sentences are extracted by searching phrases existing in the database and combining searchable phrases. In other words, assuming that phrases searched on the Internet are registered in the search engine's database, the phrases "My name is" and "HanAll" cannot be searched as connected sentences, but can be searched as individual phrases and words, so the phrases searched in this way are connected to be extracted as one prediction sentence. Therefore, "My name" + "HanAll" are connected by 'space', and "My name is HanAll" is extracted, as shown in Figure 13(a), so that "My name is HanAll" is displayed in the keyboard as a predicted sentence and can be selected. There will be. And when "My name is HanAll" is selected and entered, as shown in Figure 13(b), the symbol keyboard corresponding to the 'sym' key, the last key sequence in Equation 16, is displayed, and the period ('.') is displayed. Input can be made. If the predicted sentence shown in Figure 13(a) is not "My name is HanAll", press the central 'ret(urn)' key to proceed with the sentence registration process shown in Figure 16.

<Change the contents of the ambiguous 'sym' key to suit the search and sentence input situation and set the keyboard to divide the 'space' key into the 'sentence-space' key and the 'search-space' ('&space') key>

In Example 18, the 'sym' key was set to mean 'space' and was explained. In fact, if the 'sym' key is an ambiguous key containing all symbols as shown in Table 2, the key input shown in Equation 16 is "My name In order to efficiently search for "is HanAll.", it is desirable to distinguish between the symbol and the 'space' (more precisely, 'sentence-space') key. In order to search sentences, the character represented by the 'sym' key often corresponds to 'space' (a space character) rather than a symbol, so the arrangement of the keyboard shown in Figure 10 is changed as shown in Figure 14a to replace 'space'. When the key input of Equation 16 is made by separating the input from the symbol, 'sym' in Equation 16 means that the search can be performed efficiently by pressing the 'space' key on the keyboard shown in Figure 14a. Furthermore, when a key input is made to search a web address, when a key input is made using the keyboard shown in FIG. 14a, pressing the 'menu' key to input a symbol activates the extended keyboard shown in FIG. 14b to enter 'sym'. 'You must press the key. However, since there is no 'space' in a web address, in this case, the 'sym' key is placed instead of the 'space' key as shown in Figure 14c, so that the 'sym' key is not necessarily entered in the 'menu' when entering the key to search the web address. This eliminates the inconvenience of having to press a key.

< Extraction process of prediction sentences using group key sequence when there is no content that exactly matches the input key sequence>

The search process of Example 17 is explained by key input using the keyboard of FIG. 14A as follows. The key sequence corresponding to Equation 16 can be conceptually expressed as the following Equation 17. In Equation 17, the 'space' key is indicated by '+' and the 'sym' key is indicated by #. A, B, C, and D represent key sequences corresponding to the words 'my', 'name', 'is', and 'HanAll', respectively.

A+B+C+D# ....(17)

Key sequence comparison for searching sentences and phrases: To predict the entire sentence, first compare the key sequence of the sentence (or phrase) in the database with "A+B+C+D#" in Equation 17 to see if there is a match. If there is no match after searching, compare "A+B+C", "B+C+D", and if there is no match here, compare "A+B", "B+C", and "C+D" in order to find the best match. A large phrase is selected, the key sequences of phrases or words that do not belong to this phrase are compared and searched against the database, matches are extracted, and then combined with the content of the largest phrase already searched, the process is displayed as a predicted sentence. If there are phrases that match “A+B+C” and phrases that match “B+C+D,” the extracted sentences are determined in two ways.

(I) In the first case, what corresponds to “B+C” in the passage matching “A+B+C” corresponds to “B+C” in the passage matching “B+C+D”; If the content is the same, you can extract the predicted sentence (or phrase) by connecting the content corresponding to "A" in the phrase matching "A+B+C" to the phrase corresponding to "B+C+D". In Table 9, case 1 matches “A+B+C” in Equation 17, case 2 matches “B+C+D” in Equation 17, and in both cases, the contents corresponding to “B+C” match. In case 1 of Table 9, the content corresponding to "A" is extracted, and in case 2 of Table 9, the content corresponding to "D" is extracted, and "My name is HanAll" is extracted as a prediction sentence.

(II) In the second case, cases 2 and 3 of Table 9 are consistent with “A+B+C” in Equation 17, but “name is” and 3 correspond to “B+C” in equation 17. Since the "oboe is" corresponding to "B+C" does not match, in this case, select case 2 of Table 9 where "B+C+D" matches and enter the part corresponding to "A" in 3 of Table 9. By selecting “My” in number 1, “My name is HanAll” is extracted as a prediction sentence. In other words, if "A+B+C" matches and "B+C+D" matches, the content that matches "B+C+D" in the latter part of the key sequence is selected first, and the rest is used as a short key. The one that matches sequence group "A" is selected.

A B C D One My name is Tom 2 His name is HanAll 3 My oboe is this

When Equation 16, which corresponds to the key input of the embodiment, is keyed with the keyboard shown in FIG. 14A, the key to input 'space' is not the 'sym' key, but is input by pressing the 'space' key designated only for 'space'. Therefore, Equation 16 is changed to Equation 18. After the key input is made in Equation 18, the predicted sentence is not extracted through the search, so when the 'REG' key is pressed to proceed with the registration process, the result is shown in Figures 12(e), (g), and (i). There is no problem even if the displayed 'space' input is omitted, which has the advantage of simplifying the sentence registration process. In other words, using a keyboard with a single 'space' key shown in Figure 14a not only makes the search process more efficient, but also has the advantage of making the registration process more efficient when there is no extracted sentence after the search.

'bn m '-'t y u'- (18a)

'space'- (18b)

'b n m'-' a sdf'-'bn m '-'qw e r'- (18c)

'space'- (18d)

' i op'-'a s df'- (18e)

'space'- (18f)

'g h jkl'-' a sdf'-'b n m'-' a sdf'-'ghjk l '-'ghjk l'- (18g)

'sym' (18h)

<Configuration of a circular keyboard and a square keyboard>

The keyboard in Figure 14 has a circular shape, and its application to a square-shaped device is shown in Figure 15. And this square shape can be easily applied to the remote control-type directional keypad shown in Figure 16, so it can replace the keyboard in the remote control and fully perform the function of an input keyboard, such as writing sentences and searching the Internet through easy character input. Moreover, it brings the advantage that it can even be applied to the joy-stick of a game console. Therefore, it has the advantage of maintaining unity as an input device for wearable devices as well as circular and square mobile devices.

Figures 17 and 18 show a keyboard for predictive Korean input. Korean predictive input using this keyboard is configured to perform word and sentence prediction simultaneously.

In Figure 17, (A) is a predicted sentence display area, (B) is a keyboard key area, (C) is a currently input character display window, and (D) is a predicted word display area.

Figure 18 shows 10 key areas, which is the area (B) in Figure 17. The predictive input method using this keyboard is the same as the predictive input method of the English keyboard shown in Figures 1 and 9, and each key and the corresponding numeric code are as shown in Table 10.

ambiguous key
'ㄱㅋㄲ
ㅏㅑ
qwer
0'
'No
ㅓㅕ
tyu
One'
'ㄷㅌㄸ
ㅗㅛ
iop
2'
'sorry
ㅜㅠ
asdf
+5'
'Fuck tt
ㅡㅣㅢ
ghjkl
3'
'Shhhh
ㅐㅒ
zxcv'
'Ssssssssssssssssssssssssssd
ㅔㅖ
bnm
4' numeric code
One
2
3
4
5
6
7 ambiguous key
'grammatical space'
'space' numeric code
8
9

If you want to input “I went to Seoul,” just press the key corresponding to each letter according to the configuration in Table 10, and the key input sequence is as shown in equation (19) below.

In Seoul: ⑥②④④②-⑧-④⑦- (19a)

<Space>: ⑨- (19b)

Went: ①①⑥-⑧-④②④③ (19c)

In other words, the first phrase of 'I went to Seoul', 'Seoul', is a phonetic part containing the particle 'E', and to input it, enter the key composition corresponding to the particle 'Seoul' and the particle 'E' in order as shown below. Just do it. Enter the numeric code '⑥②④④②' corresponding to 'Seoul' and the numeric code '④⑦' corresponding to the particle 'E', but if you press the 'grammatical space' key '⑧' between them, '⑧' Predicted words corresponding to '⑥②④④②' are extracted from the 'noun, pronoun, rhetoric' group in the dictionary database, such as 'Seoul', 'skin', 'preface', 'facade', 'seoul', ' ‘Seoun’, ‘Seoyun’, etc. are extracted as predicted words. Next, the predicted word 'e' corresponding to the key input sequence of the particle '④⑦' was extracted from the 'posal' group of the dictionary database, and these vowel prediction words and the particle prediction word were combined to make 'to Seoul', ‘Heomul-eh’, ‘Seomun-eh’, ‘Heomul-eh’, ‘Seomul-eh’, ‘Seowoon-eh’, and ‘Seoyun-eh’ are displayed in the predicted word display area as integrated prediction words (precisely prediction ‘gu’). . Therefore, you can select and input the one you want to input among these, and in the predictive input system, 'To Seoul', which is extracted as the first predicted word, is temporarily entered in the sentence input window (Figure 17-(B)). . If 'in the preface' extracted as a predicted word is the word you want to input, and if the input device is a touch screen, touch the predicted word input window area where 'in the preface' is displayed in Figure 17 or enter the corresponding word predetermined on the input device. By executing the method, the default prediction word is displayed in place of 'Seoul' displayed in the input window.

As seen above, if the words corresponding to the proverb and the words corresponding to the particle are classified into different groups and extracted as predicted words, the key '⑧' corresponding to the grammatical space is shown in equation (19a). The key input on the front is used to extract the vowel word, and the key input on the back of key '⑧' is used on the extract of the particle word.

Like verbs, in the case of verbs that involve the conjugation of the endings of verbs or adjectives, the ending of the verb corresponding to the particle of the verb and the stem of the verb corresponding to the verb are entered into the dictionary database as 'end' group and 'stem' group, respectively. By constructing (group), the key order before and after the 'grammatical blank key' is applied to extract stem words and word endings from these groups. In other words, stem prediction words extracted according to the key input sequence before the 'grammatical space key' and extracted according to the key input sequence before the 'grammatical space key' from a database separated into term stem group and ending group. You can combine predicted word endings and display them in the predicted word input window as a verb prediction word (more accurately, a prediction phrase) and select/enter it. To explain Equation (19c) as an example, 'wen', which corresponds to the verb 'went to Seoul', is composed of 'gan', which corresponds to the stem, and 'eo', which corresponds to the ending, so the key input order is 'gan', which is the stem. '①①⑥' corresponding to ' and '④②④③' corresponding to the ending 'eo' are connected by '⑧', a grammatical space symbol. Therefore, using grammatical blank keys, the noun and particle are distinguished in the case of the verb, and the verb is registered as the noun group, particle group, (verb) stem group, and (verb) ending group in the dictionary database by distinguishing the stem and the ending. This provides a way to drastically reduce the number of words (to be precise, 'phrases') corresponding to the verb part created by combining the particle and the particle and the verb part created by combining the stem and the ending of the verb. To explain this specifically, assuming that the number of pronouns in the dialect part is 100,000 and the number of particle phrases that can be combined with the dialect phrase is 1,000, the number of dialect phrases combining the pronoun and particle amount to 100 million. In the case of verbs, if the number of verb stems is 100,000 and the number of endings that can be combined with the stem is 1,000, the number of verb phrases combining stems and endings also reaches 100 million. However, the number of endings that can be combined with the actual stem is simply '-da'('ga'+'da'->'ga-da'), '- is'('ga'+' is'->'ga-gan') ) to '-jimanseoyo' ('ga'+'jimaneodo'->'eggplantmandoyo'), the number of cases is well over 100,000, so the number of verbs combining these endings with the stem is 1 billion. It goes beyond dogs. Therefore, when registering the stem and ending of an verb separately in the dictionary database, 200,000, which is the number of stems plus 100,000, need to be registered in the dictionary database, so that all verbs combining stems and endings can be registered. The capacity is reduced by approximately 50,000 times compared to the previous case. This reduced capacity is not only a problem due to the difference in size, but in reality, the time to search the database to extract predicted words (phrases) from the database differs by more than 10,000 times, making it impossible to compare in terms of practicality.

In the case of English, when forming a phrase, the phrase equivalent to 'Seoul' in Hangul is expressed as 'To Seoul', so the preposition 'To' that creates the phrase and 'Seoul' corresponding to the phrase are spaces, so 'To' Since 'Seoul' is distinguished from 'Seoul', there is no need to predict speech parts that combine prepositions and pronouns, so the words registered in the database are the sum of the number of prepositions and the number of pronouns (nouns). Therefore, even if the number of prepositions is as large as the number of nouns, the number of registered words increases by about two times. On the other hand, in Hangul, a particle (corresponding to a preposition in English) and a particle (corresponding to a noun in English) are combined, so the 'grammatical space' of the present invention is intentionally inserted between the particle and the particle (in the case of a verb, the stem and the ending). This dramatically reduces the number of words that need to be registered in a dictionary database, such as in English, to prevent the storage capacity of the dictionary from increasing, and further reduces the number of registered words, which has the effect of dramatically reducing search time. .

The Hangul predictive input method using the 'grammatical space' key examined in Example 22 shows how the database of Hangul word phrases (verb phrases, verb phrases) can be dramatically reduced. In order to use the 'grammatical space' key that brings these advantages, the user must have a grammatical understanding that can clearly distinguish between nouns and particles and the stems and endings of verbs. However, in the case of 'to Seoul', it is easy to distinguish between 'Seoul' and 'e', but in the case of the verb 'went', it is not easy to distinguish whether it is 'went' + 'yo' or 'went' + 'yo'. User confusion may occur. Even in the case of ‘I went there’, it becomes even more difficult to distinguish. To overcome this problem, in this embodiment, 'grammatical space' is inserted between each syllable of Hangul, so that word phrases (verb phrases and verb phrases) are retrieved from the dictionary database even if the user does not recognize the actual location of the 'grammatical space' key. Shows how to extract it.

Taking 'I went to Seoul' as an example, 'To Seoul', which corresponds to a declarative phrase, is easy to distinguish between an adjective and a particle, so 'I went', which corresponds to an verb phrase, is selected and the key input sequence is as in Example 22. corresponds to equation (19c). However, if a 'grammatical space' key is inserted for each syllable according to the design of this embodiment, the key input sequence becomes Equation (20) below.

Went: ①①⑥-⑧-④②-⑧-④③ (20)

When the key input sequence as in equation (20) is achieved, the method of extracting verbs from the database is to extract predicted stem words and ending words corresponding to the key input sequence before and after each 'grammatical space' key from the stem group and the ending group, respectively. It is to combine. In other words, the key input sequence in equation (20) is converted into the following two equations (21a) and (21b) during the prediction word extraction process, and the stem word and suffix word corresponding to the key input sequence before and after the 'grammatical space' key in each equation. goes through an extraction process. To explain this in light of Table 23, which shows the prediction process for word phrases containing grammatical spaces, the process of predicting phrases and verb phrases according to the key input sequence containing grammatical spaces is performed for grammatical spaces inserted between syllables to make predictions. The phrase is extracted. In the case of equation (20) above, the grammatical space utilization word phrase prediction process shown in Table 23 is performed twice. In this way, in order to predict 'went', a grammatical space is entered between each syllable rather than entering a grammatical space exactly once between 'went' and 'yes', and the process in Table 23 is performed one more time. However, even if the user does not know exactly where to input the grammatical space, it has the advantage of being able to input the word phrase to be input by extracting it from the dictionary database.

I went : ①①⑥-⑧-④②④③ (21a)

I went : ①①⑥④②-⑧-④③ (21b)

Went: ①①⑥④②④③ (21c)

In the extraction process using the key input sequence corresponding to equation (21a), 'went' is extracted, but in the case of equation (21b), the verb part is not extracted because the word 'went' is not registered in the stem group of the database. Because it doesn't. Therefore, when the 'grammatical space' key is inserted between each syllable, the user has the advantage of being able to predict the desired word phrase without having to worry about grammar.

In addition, in the case of Korean, there may exist words that do not require grammatical spaces that do not combine with particles or endings, as shown in equation (21c), so independent words in addition to word phrases containing particles or endings must be searched in the dictionary database, so Eq. In the case of the key input sequence corresponding to (21c), separately from Table 23, Korean word prediction input must be performed in parallel like English word prediction input, and independent word groups for this must be included in the dictionary database. And the extraction of these independent words is the same as the English word prediction method, and in Figure 23, the process is the same as the extraction of the word list from the chair group. The independent words extracted in this way are combined with the list of phrases and phrases and included in the entire word (phrase) prediction list, so that the user can select/enter them.

This example explains how to apply the 'grammatical space' inserted between each syllable shown in Example 23 above to extract syllables of Hangul. As already seen in Example 22, the dictionary database for predictive input of Hangul has the problem that the size of the word phrase itself becomes too large to register, so word phrases are divided into noun groups and particle groups (in the case of verbs, stem groups and (separated into parent groups) were registered. However, the problem is that we need to provide a way to input words in cases where they are not registered in these databases. However, registering unusual words such as people's names and place names would be too large, so to overcome this problem, they are not registered together with names. It provides a method for users to combine syllables of unknown words by extracting them one by one.

The key input sequence for 'I went to Seoul' according to the configuration of this embodiment is the same as equation (22). When the key input corresponding to 'to Seoul' is made according to equation (22), the syllable is predicted as shown in Figure 22. The extracted syllables are displayed in the window, and 'seo', 'heo', 'seo', 'tongue', and 'she' corresponding to the key input '⑥②' of the first syllable in equation (22a) are extracted, and the second syllable is extracted. 'Moon', 'Water', 'Woon', 'Ul', 'Wo', 'What', 'Yun', 'Yul', and 'Mule' corresponding to the syllable key input '④④②' are extracted, and the last 'E', 'Yes', 'Me', and 'ㅞ' corresponding to the syllable key input '④⑦' are extracted and displayed, so that the user can easily select them by arranging them in order from the highest frequency of each syllable. In other words, in the case of the syllable shown in Figure 22(E), looking at the predicted syllable corresponding to the second syllable, the frequency of 'moon' is greater than 'ul', so 'moon' is arranged before 'ul'. The order of the first letters of the extracted syllables is displayed as 'seo', 'moon', and 'e', that is, it is predicted differently from 'Seoul', which is the first of the predicted word phrases shown in Figure 22-(D). This can be seen because the arrangement order of the predicted word phrases is also arranged according to the frequency of words (verbs or verbs) in the phrase, so the frequency of use of 'Seoul' is greater than that of 'preface'. ', 'In the preface'.

‘In Seoul’: ⑥②-⑧-④④②-⑧-④⑦- (22a)

<Space>: ⑨- (22b)

‘I went’: ①①⑥-⑧-④②-⑧-④③ (22c)

In the case of Hangul, the number of usable syllables is 11,172, so the number of commonly used verb stems amounts to hundreds of thousands (the number of vocabulary included in the Standard Dictionary of the Korean Language published by the National Institute of the Korean Language in 1999 is approximately 500,000). As shown in equation (22), syllable prediction is as easy as word prediction for key inputs separated by 'grammatical space' keys, so it functions as a complement to the word prediction input method so that words that are not registered in the database can be entered. do. As a result, by entering the 'grammatical space' key between each syllable, as described in Example 23, you can easily use word prediction input without knowing the grammatical structure of the word phrase, and you can enter words that are not registered in the dictionary database. This brings the advantage of implementing all possible methods simultaneously.

The above shows that word prediction input in Hangul can be easily implemented using the 'grammatical space' key. This implementation of word prediction input in Hangul can further increase input efficiency by enabling sentence prediction input as in English.

The Korean sentence prediction input method is applied in the same way as the English sentence prediction input method. This is a method that allows sentences to be predicted even if only part of each word group is entered in the key input sequence separated by space keys. Figures 20 and 21 show that in the key input sequence for 'I went to Seoul', if part of the key input sequence corresponding to 'To Seoul' and part of the key input sequence corresponding to 'went' are entered, 'I went to Seoul' is entered. It is extracted from the sentence database and displayed in the ‘predicted sentence display window’ (Figure 20(A), Figure 21(A)).

Figure 20 shows that even when a key input corresponding to 'I went to Seoul' is made, 'I went to Seoul' is extracted from the database and displayed in the 'predicted sentence display area', and Figure 21 shows the first letter of each word phrase of 'I went to Seoul'. Even if a key input corresponding to 'I went to Seoul' is made, 'I went to Seoul' is extracted from the database and displayed in the 'predicted sentence display area'.

Table 11 shows the key sequence for ‘I went to Seoul’ registered in the dictionary database. This table shows two key input sequences. Method 1 is a key input sequence that includes 'grammatical spaces' in the word group, and Method 2 is a key input sequence that does not include 'grammatical spaces'. There is no problem in the Korean sentence prediction of the present invention even if the key input sequence is set without the 'grammatical space' key as shown in Method 2 in Table 11. The reason is that in order to extract 'I went to Seoul' as a sentence, the word phrase corresponding to 'to Seoul' is already specified in the dictionary database, so there is no need for the 'grammatical space' key. Therefore, unlike word phrase prediction, which extracts and combines nouns and particles separately to reduce the number of word phrases, the method of registering sentences in a dictionary database for sentence prediction input is according to Method-1 or Method-2 in Table 11. You can register the key input sequence of the sentence. Even if the dictionary database for sentence prediction uses Method-2 in Table 11 for sentences and key input sequences as shown in key sequence-A in Table 12, the key input sequence entered by the user during the sentence extraction process You can use it after removing the 'grammatical space' and converting it to key sequence-B in Table 12.

The key input sequence in Tables 12 and 13 allows sentences to be extracted from the dictionary database in light of the standards for sentence prediction (Table 5) and input even if the first part of the phrase that makes up the sentence is entered. That is, Table 12 and Table 13 are key input sequences corresponding to the input situations shown in Figures 20 and 21, respectively. It shows that the key input sequence corresponding to each part of the phrase 'I went to Seoul' has been established in the database for Korean sentence prediction input.

sentence In Seoul gap went key sequence Method-1 ⑥②④④②-⑧-④⑦ ⑨ ①①⑥-⑧-④②④③ Method-2 ⑥②④④②④⑦ ⑨ ①①⑥④②④③

sentence first key sequence gap second
key sequence key sequence-A ⑥②④④②-⑧-④⑦ ⑨ ①①⑥ key sequence-B ⑥②④④②④⑦ ⑨ ①①⑥

sentence first key sequence gap second
key sequence key sequence-A ⑥② ⑨ ①①⑥

This embodiment teaches a method of searching the Hangul database structure (verb phrase, verb phrase) shown in FIG. 23 without using grammatical space keys. In other words, in the case of Korean, unlike English, there are changes in particles and endings at the end of nouns and verbs, so there are too many changes that each noun and verb can have, so in order to reduce the database, it is necessary to divide nouns and verbs into particles and endings. This is the core of Figure 23. That is, in cases where the 'grammatical space' introduced in Example 22 is unfamiliar to users and they do not know exactly where the grammatical space should be specified in the actual phrase, they must cumbersomely enter the key corresponding to the grammatical space for each syllable. In either case, it can be inconvenient for the user. Therefore, we would like to provide a method of searching the Korean database shown in Figure 23 without entering grammatical space keys.

If the user enters the key sequence-B in Table 12 without a key corresponding to a grammatical space to input 'I went to Seoul', the method of searching the Korean database proceeds as shown in Table 14. That is, assuming that the input key sequence '⑥②④④②④⑦' is followed by a grammatical space for each key, it is divided into key inputs corresponding to verb groups and particle endings, corresponding to search orders 1 to 8 in Table 14. The process involves searching the database and extracting phrases by combining the predicted words corresponding to the verb groups and particle suffixes obtained from the database.

Search order Cheeon-gun (Yongeon-gun) particle (final) One ⑥②④④②④⑦ 2 ⑥②④④②④ ⑦ 3 ⑥②④④② ④⑦ 4 ⑥②④④ ②④⑦ 5 ⑥②④ ④②④⑦ 6 ⑥② ④④②④⑦ 7 ⑥ ②④④②④⑦ 8 ⑥②④④②④⑦

If there is no word in the database that corresponds to any one of the noun phrases or particle endings, it is assumed that there is no predicted word. For example, in the case of search order 3 in Table 14, even if a word matching the key input sequence '⑥②④④②' corresponding to the verb group (verb group) exists in the database, the key input sequence corresponding to the particle (final) is '④⑦'. If the correct word does not exist in the particle (final) part database, it is determined that there is no verb for search order 3, and only when a word that matches both the verb group and the particle (final) exists, the verb group is selected. The extracted words of (verb group) and the extracted words of particle (ending) are combined to extract a phrase corresponding to a phrase (or verb).

In general, the number of words registered in the database is greater in the number of word groups (verb groups) than the corresponding particles (suffixes), so one of the ways to quickly proceed with the search order shown in Table 14 is to use the particles and endings corresponding to the particles. First, search the database for the key input sequence to see whether there is a (particle or suffix) word that matches the key input sequence. If the correct (particle or suffix) word does not exist, search for the word group (verb group). Skip the search. In this way, the entire search process can be performed efficiently by reducing the search process.

Japanese has a grammatically similar structure to Korean, except that there are no spaces. Particles are attached to the verb part to make its meaning clear, and the endings of the verb part are varied so that one verb can take on numerous forms. Therefore, unlike English, the contents of Examples 22-26 can be applied to Japanese to reduce the database according to changes in particle and verb endings. However, in the case of Japanese, there is no space in the sentence, so if a space is given arbitrarily like in Hangul, word prediction input can be easily applied by applying the contents of Examples 22-26 to build a database of verb groups and particles (final endings). Furthermore, the sentence prediction input of the present invention can also be applied.

division sentence post office verb part (One) I am a student I am I'm a student (2) わたしはがくせいです
(private life) (3) わたしは_がくせいです わたしは がくせいです (4) watashiwagakuseidesu (5) watashiwa gakuseidesu watashiwa gakuseidesu

In this regard, the idea of this embodiment is to apply space characters like Korean characters to implement word prediction input in Japanese. Table 15 shows ‘I am a student’ in Korean and Japanese. (1) in Table 15 is the Korean sentence 'I am a student', with the subject 'I' and the verb 'I am a student' separated by a space. In contrast, in Japanese, " " is equivalent to 'I am', 'private' is equivalent to 'I am a student'" " There are no spaces between them. And for Japanese prediction input, when inputting in romanji, an English expression of Japanese, it is expressed as shown in (4) of Table 15. In this too, there are no spaces. However, this If a grammatical space is inserted according to the configuration of the embodiment, it is expressed as in (5) of Table 15. In this case, the speech part and the verb part are separated, and 'watashiwa' corresponding to the verb part is separated from 'gakuseidesu' corresponding to the verb part. The method of extracting from the database is to adopt the method of Example 26. That is, in the database, 'watashi (私)', which corresponds to 'I' in Korean, is registered in the vowel group, and the particle 'eun' is registered in the database. The 'は' is registered in the survey department, so it is extracted as a predicted word corresponding to each speech group and particle, and then combined and input as a predicted word (phrase) in the speech group, and the romanji expression shown as (4) in Table 15. For predictive input of a sentence, the keyboard shown in Figure 9(a) can be used, and for the predictive input method of the present invention, the criteria shown in Table 5 can be applied for the predictive input method and sentence predictive input.

Until now, the sentence prediction input method was a process for inputting actual sentences. However, the sentence prediction input method of the present invention can be applied even when it does not have a grammatical structure. For example, when selecting a song from a music streaming service or a movie from a video streaming service, in the case of music, in addition to the song name, the name of the composer, lyricist, performer, singer, etc. is used to search. When selecting a song like this, If each song is divided as shown in Table 16, in order to select the song, you must enter the name of the song, composer, and performer separately to select the desired song. For each song in Table 16, the key sequence for the classification content (song title, composer, performer) is displayed.

No sentence key sequence One Nocturne Chopin Seoul Symphony Orchestra 41614131-9-63564-9-624426513614411312 2 Nocturne Chopin Berlin Symphony Orchestra 41614131-9-63564-9-5725225213614411312 3 Moonlight Sonata Beethoven Seoul Symphony Orchestra 44221314632131-9-5733572-624426513614411312 4 Moonlight Sonata Beethoven Berlin Symphony Orchestra 44221314632131-9-5733572-5725225213614411312

As shown in Table 16, if the song list is in a database, in order for the user to extract song number 1 (Chopin's Nocturne, performed by the Seoul Symphony Orchestra) from this database, the sentence corresponding to 'Chopin's Nocturne, Seoul Symphony Orchestra' must be entered using the sentence prediction input method. The key sequence that must be entered is the key sequence of song 1 in Table 16, and the key input sequence that the user must actually input is shown in Equation 23. Enter the key arrangement '④①⑥①④①③①' corresponding to the name of the song 'Nocturne', enter the space key '⑨' to separate the items of the song, and then separate the items with the key arrangement '⑥③⑤⑥④' corresponding to the composer 'Chopin'. If you then enter the blank key '⑨' and finally enter the key arrangement '⑥②④④②⑥⑤①③⑥①④④①①③①②' corresponding to the performer 'Seoul Symphony Orchestra', the standard in Table 5 (the ambiguous key designation corresponding to each character follows the contents of Table 10) Based on this, song number 1 is selected and extracted from the song list database in Table 16 using the sentence prediction input method.

Nocturne: ④①⑥①④①③① (23a)

gap : ⑨ (23b)

Chopin: ⑥③⑤⑥④ (23c)

gap : ⑨ (23d)

Seoul Symphony Orchestra: ⑥②④④②⑥⑤①③⑥①④④①①③①② (23e)

In addition, in the sentence prediction input method performed according to the standards in Table 5 of the present invention, song number 1 in Table 16 can be extracted even if a key input corresponding to a portion of the phrase constituting the sentence is made as shown in Equation 24. In other words, a key input corresponding to the (front) part of each item (song title, composer, performer) that makes up song number 1 in Table 16 is made, and song number 1 in Table 16 is extracted. In other words, even when selecting something other than a sentence such as music or a movie through a key input such as Equation 24, the sentence prediction input method of the present invention provides a way to easily select music and a movie. In Table 16, each song is displayed as a sentence, and the items of each song (song title, composer, performer) are separated by 'spaces' and can be extracted using the sentence prediction input method of the present invention, and the abbreviated key input as shown in Equation 24 It can be easily extracted.

Nocturne: ④① (24a)

gap : ⑨ (24b)

Chopin: ⑥③ (24c)

gap : ⑨ (24d)

Seoul Symphony Orchestra: ⑥②④④② (24e)

In the case of sentences, the order of the phrases that make up the sentence is determined by the grammatical structure. For example, if you say, ‘I go to school,’ do not say, ‘I go to school.’ Therefore, when registering the sentence 'I go to school' in the database of the sentence prediction input method of the present invention, the key arrangement corresponding to 'I go to school' is entered together with this sentence in the form shown in Table 16. You will be registered. However, the key arrangement that users enter to select music or movies in streaming services is as shown in Table 16. To select song number 1, you can key in 'Chopin Nocturne Seoul Symphony Orchestra', but you can key in 'Chopin Nocturne Seoul Symphony Orchestra' You can also enter '. If the composer is entered first like this, song number 1 cannot be extracted if the database is structured as shown in Table 16. In this regard, in the process of selecting a music song or movie, as shown in Table 17 for song number 1 in Table 16, so that users can select the song they want even if they enter 'song title, composer, and performer' in any order. In addition to the key sequence corresponding to the original sentence 1-a, five additional key sequences, including 1-b, 1-c, 1-d, 1-e, and 1-f, were added to the key sequence corresponding to song 1 in Table 16. Add and register in the database. In fact, the sentences that these key sequences 1-a to 1-f mean are the sentences shown in parentheses under each key sequence in Table 17. However, the actual database registered as a sentence corresponding to each key sequence is 'Nocturne Chopin Seoul Symphony Orchestra' in all key sequences from 1-a to 1-f. Therefore, when the user enters the key arrangement in 1-a to 1-f, in the process of extracting the sentence from the database, 'Nocturne Chopin Seoul Symphony Orchestra' is extracted, so that the song actually selected is song number 1 in Table 16. will be.

No key sequence sentence 1-a 41614131-9-63564-9-624426513614411312 Nocturne Chopin Seoul Symphony Orchestra 1-b 41614131-9-624426513614411312-9-63564
[Equivalent to ‘Nocturne Seoul Symphony Orchestra Chopin’] Nocturne Chopin Seoul Symphony Orchestra 1-c 63564-9-41614131-9-624426513614411312
[Corresponding to ‘Chopin Nocturne Seoul Symphony Orchestra’] Nocturne Chopin Seoul Symphony Orchestra 1-d 63564-9-624426513614411312-9-41614131
[Corresponding to ‘Chopin Seoul Symphony Orchestra Nocturne’] Nocturne Chopin Seoul Symphony Orchestra 1-e 624426513614411312-9-41614131-9-63564
[Equivalent to ‘Seoul Symphony Orchestra Nocturne Chopin’] Nocturne Chopin Seoul Symphony Orchestra 1-f 624426513614411312-9-63564-9-41614131
[Corresponding to ‘Seoul Symphony Orchestra Chopin Nocturne’] Nocturne Chopin Seoul Symphony Orchestra

Therefore, when registering an expanded key sequence for the same sentence (song) in the sentence database of Table 16 as shown in Table 17, users can easily select the same song by keying in without distinguishing the song name, composer, or performer. . In other words, as shown in equation (25), even if the user inputs a key, it matches the key arrangement corresponding to 1-c in Table 17, so song 1-c (actually the selected song is the same as song 1 in Table 16) is extracted. This results in the same result as extracting song 1 in Table 16.

Chopin: ⑥③ (25a)

gap : ⑨ (25b)

Nocturne: ④① (25c)

gap : ⑨ (25d)

Seoul Symphony Orchestra: ⑥②④④② (25e)

In Example 29, all cases in which the phrases constituting the sentence can be placed in the database by changing the order of the phrases within the sentence so that the sentence can be extracted from the character prediction input regardless of the order of the phrases constituting the sentence Registered. However, in this embodiment, as shown in Table 17, a database containing a list of key sequences containing the order of all phrases constituting the sentence is not used, and only one key sequence corresponding to one sentence is registered as shown in Table 16. Even if a database is used, we explain how to extract sentences in which the order of phrases does not change even if the user changes the order of phrases in the sentence, as shown in Equation 25. In other words, this explains how to extract sentence 1 in Table 16 even if the user inputs a key as shown in Equation 25.

The contents of Table 5, which is the standard for the sentence prediction input method so far, checks whether the user's input key arrangement and the key arrangement of the sentence registered in the database match from the first key arrangement of the sentence, and if they match, the key arrangement registered in the database is checked. Extract sentences. In contrast, in this embodiment, the key arrangement entered by the user is compared with the key arrangement of the sentence registered in the database, regardless of the order of the phrases constituting the sentence, and if any phrase of this sentence matches the corresponding key arrangement, the key arrangement is compared. Sentences are extracted. This is explained using Equation 25 as an example, and the first key sequence entered in Equation 25 is '63' in Equation 25a. Until now, only matching sentences were extracted by comparing this key array '63' with the key array corresponding to the first phrase of all sentences in the database, but in this embodiment, not only the first phrase but also all phrases of the sentences registered in the database were extracted. Compare the key arrangement with the user's first phrase key arrangement '63' to check if there is a matching key arrangement. If there is a matching sentence during this process, it forms a temporary sentence list for the next search process. This is the first process. (Step 1) In the first process, if there is a temporary sentence list, the key array '41' corresponding to the second phrase entered by the user is searched in the same way as the first process. . If there are matching sentences in this process, these sentences are reconstructed into a temporary sentence list and proceed to the next step. This process is the second step. (Step 2) Also, a temporary sentence is created for the last key sequence '62442' entered by the user. Search the list and extract matching sentences. This process is the third step. Assuming that Table 16 is a database, Step 1 to Step 3 above are sequentially explained as follows.

< Step 1 >

Among sentences 1 to 4 in Table 16, sentences 1 and 2 have a key arrangement that matches the key arrangement '63' (Equation 25a) entered by the user. In other words, Sentence 1 and Sentence 2 match because the key sequence corresponding to the second phrase of these sentences is '63564', which starts with '63'. (Here, the match does not mean that the entire part is the same, but it is judged to match if the input key array is compared from the beginning and the key array corresponding to the same number of keys from the beginning of the key array of the relevant passage in the database is the same. That is, Table 18 Since the key arrangement corresponding to the first phrase of sentences 1 and 2 is '41614131', the first two keys are the same as '41', so a match decision is made.) For the next step search with sentences 1 and 2 in Table 16 A list of search sentences is formed.

< Step 2 >

Search whether the second key array '41' (Equation 25c) entered by the user matches Sentence 1 and Sentence 2 in Table 16, which is the second-stage search sentence list. The key sequence corresponding to the first phrase of these sentences is '41614131', which matches because it starts with '41'. Therefore, Sentence 1 and Sentence 2 in Table 16 become the search sentence list for the next step.

< Step 3 >

Search whether the third key arrangement '⑥②④④②' (Equation 25e) entered by the user matches Sentence 1 and Sentence 2 in Table 16, which is the 3-step search sentence list. Among these sentences, in the case of sentence 1, the key sequence corresponding to the third phrase is '624426513614411312', which matches because it starts with '⑥②④④②'. On the other hand, in the case of sentence 2, the key sequence corresponding to the third phrase is '5725225213614411312' and does not start with '⑥②④④②', so it does not match. Therefore, sentence 1 in Table 16 is selected as the extracted sentence.

Looking at the example shown above, the standards for the sentence prediction input method shown in Table 5 can be summarized as Table 18. In particular, for additional explanation regarding criterion 2 in Table 18, when the key arrangement entered by the user is the same as equation (26), the process of extracting sentences from the database in Table 16 is as follows. First, among the key arrays entered by the user, the longest segment separated by spaces is '416' in equation (26c). Therefore, comparing '416' with the database in Table 16, Sentence 1 and Sentence 2 match in the first phrase, so Sentence 1 and Sentence 2 become the search sentence list for the next step. The second search step compares the key array '63' in equation (26a) with the second-step search sentence list, Sentence 1 and Sentence 2. Since it matches the first part of the second phrase of these sentences, Sentence 1 and Sentence 2 are The next step is the list of search sentences. Next, comparing the key arrangement '41' of Equation (26e) with Sentence 1 and Sentence 2, which are the three-step search sentence list, it matches the first part of the first phrase of these sentences, but this phrase has already been changed to the key arrangement of Equation (26c) in the first step. Since it matches with , the match is judged as a duplicate, so the user key input shown in equation (26) does not have a matching sentence. In other words, the key arrangement in equation (26e) is '41', but this is a key input by the performer corresponding to 'Ashkenazy', so a matching sentence (song) cannot be found in the database in Table 16.

Standard 1
The number of segments separated by blank keys in the key array entered by the user is
It must be equal to or smaller than the number of segments constituting the sentence registered in the database.
Standard 2
When criterion 1 is met,

In the key array entered by the user, the number code sequence corresponding to the segment separated by a blank key must match the number code sequence corresponding to the passage for at least one phrase among the sentences registered in the database by comparing it from the front. .

but,
In the key array entered by the user, segments separated by blank keys must match different passages of sentences registered in the database.
In other words, the phrases (segments) separated by blank keys in the key array entered by the user must not overlap and match the same phrases in the sentences registered in the database.

Chopin: ⑥③ (26a)

gap : ⑨ (26b)

Nocturne: ④①⑥ (26c)

gap : ⑨ (26d)

Ashkenazy: ④① (26e)

As shown in steps 1 to 3 above, we looked at a search method that can extract the desired sentence from the database even if the order of the phrases that make up the sentence is changed using Table 16. In other words, without enlarging and converting Figure 16 to Figure 17, it is possible to extract sentences from the database even if the search process is different and the user inputs the phrases that make up the sentence in a different order. In other words, when entering the selection elements (song title, composer, performer) required to select music (or a movie, etc.), the order is not important, so the elements that specify the piece (song title + composer + performer) must be combined in one sentence. Considering that the user cannot know in what order these elements are arranged in the database, this embodiment and embodiment 29 provide a method of applying the sentence prediction input method to these non-sentential elements.

In addition, when executing this embodiment and Example 29, when the number of sentences registered in the database is small as shown in FIG. 16, sentence 1 and sentence 2 are extracted and can be easily selected by entering only Equation 25a. There is. Comparing Figures 16 and 17, Figure 16 can organize the database more concisely than Figure 17, but the search process may be longer because key inputs entered by the user must be compared for all phrases during the search process. On the other hand, in Figure 17, the search process can be simplified, but the size of the database increases, so the better option should be selected in the actual application process.

In Examples 29 and 30, in order to select songs or movies that are not in grammatical sentence form, the selection elements (title, composer, performer, etc.) are composed in sentence form and extracted from the database using a sentence prediction input method and input. We looked at the case. In these embodiments, each element (song title, composer, performer) constituting the database was separated by a space key and expressed as a single phrase in the sentence. However, there may be cases where the song title, composer, and performer are not a single phrase and contain spaces. For example, in the case of 'Chopin' as the composer, it may be displayed as 'Frederic Chopin', and if this is displayed as song number 5 in Table 19, even if the key input shown in equation (25) is made, the method shown in Example 30 When searching the database represented in Table 19, song number 5 along with song number 1 can be extracted and selected. In this respect, the elements that make up sentences (songs) registered in the database are not 'song name', 'composer', or 'performer', but the details of each component separated by spaces become the elements that make up the actual database. Therefore, Table 19 shows 'song title', 'composer', and 'performer' in sentences representing the song in order to extract the song or movie displayed as a sentence into the database when displaying ungrammatical things such as music songs or movies in the form of sentences. There is no need to distinguish between the 'space' used to distinguish and the 'space' included in the song title itself. Not only the name of the composer, but also the name of the song was expanded to say 'Nocturne in C minor' instead of 'Nocturne' (Table 19 shows 'C minor' as the name of song number 6, and 'C minor' for the convenience of inputting using only the Korean keyboard. Even if the key input shown in equation (25) is made, song number 6 can be extracted and selected. In this regard, if song number 5 is registered in Table 19, there is no problem even if song number 1 is deleted. However, in the case of piece 6, a portion of Chopin's Nocturne is specified, so pieces 1 and 6 can be registered at the same time to be distinguished. Therefore, as shown in Table 19, the blank key in the key arrangement representing the song is used not only to distinguish the components (song name, composer, performer), but also for spacing within each component to form the key arrangement of the database. If you do so, you will be able to select a song by extracting sentences from this database using the sentence prediction input method based on the standards shown in Table 18.

No sentence key sequence One Nocturne Chopin Seoul Symphony Orchestra 41614131-9-63564-9-624426513614411312 2 Nocturne Chopin Berlin Symphony Orchestra 41614131-9-63564-9-5725225213614411312 3 Moonlight Sonata Beethoven Seoul Symphony Orchestra 44221314632131-9-5733572-624426513614411312 4 Moonlight Sonata Beethoven Berlin Symphony Orchestra 44221314632131-9-5733572-5725225213614411312 5 Nocturne Frederic Chopin Seoul Symphony Orchestra 41614131-9-552737251-9-63564-9-624426513614411312 6. Nocturne in C minor Frédéric Chopin Seoul Symphony Orchestra 41614131-9-552737251-9-63564-9-624426513614411312

Table 19 shows that the sentences that make up the database, such as sentence 6, can be applied to general sentences as well. In other words, regardless of whether the sentence components separated by spaces are originally the name of the song or the composer, all phrases can be extracted and selected from the sentences registered in the database by comparing them with the key arrangement entered by the user as a single element and judging only whether they match. It has been confirmed that there is. In this sense, if the criteria in Table 18 are applied to general sentences, they can be used for general searches such as Internet searches, rather than just selecting music songs or movies.

In this regard, the sentence prediction input method of the present invention applies the sentence prediction input method to Internet search, so that the Internet can be easily searched even with a keyboard having an ambiguous key as shown in FIGS. 10A and 18. For example, when the sentence 'I go to school' is entered as shown in equation (27) using the predictive input keyboard shown in Table 18, a search engine (e.g. 'Google' or Microsoft's 'Bing') As shown in Figure 24, sites containing the topic sentence 'I go to school' are extracted and shown. If the query index corresponding to the topic sentence of each of these sites is defined as an ambiguous key sequence as shown in Table 20, when a key input is made by the user as in Equation (27), sentences 1, 2, and 3 in Table 20 Since it conforms to Table 10, which is the standard for this sentence prediction input method, it is extracted from the search engine database structured as shown in Table 20 and allows selective input.

As seen above, when a user enters a key with a keyboard consisting of an ambiguous key as shown in Figure 18, matching sentences are extracted from the index database of the search engine consisting of Table 20, or a website containing these sentences is extracted. can also be extracted. In this respect, since almost all commonly used sentences exist on the Internet, it is the most ideal database for sentence prediction input of the present invention. The reason is that the inconvenience in using the predictive input method is when words, phrases, or sentences are not registered in the database. To solve this, register words, phrases, or sentences that are not registered in the database as in Example 5. It is inconvenient to have to perform a separate process during the input process. However, since almost all sentences commonly used by people are registered in the search index database of Internet search engines, the input process through a separate registration process as in Example 5 can be avoided, so sentence input through predictive input can be done using a full keyboard. It can have the same effect as inputting a sentence.

I am : ②①②⑤② (27a)

gap : ⑨ (27b)

at school : ⑥①①①③④⑦ (27c)

gap : ⑨ (27d)

going : ①①⑥②⑤③① (27e)

No website keyword sentence Search index key sequence One I am eight years old and go to school. 21252-9-423225-9-612-9-6111347-9-1162531 2 I wear gloves to school. 21252-9-714115452-9-1513-6111347-9-1162531 3 I go to school. 21252-9-6111347-9-1162531

One of the advantages of the sentence prediction input method of the present invention is that a sentence can be predicted and entered even if only a part of each word constituting the sentence is input. In addition, if you apply the Hangul initial consonant input method to the sentence prediction input method, sentence input will become easier. Therefore, as shown in No. 2 of Table 21, when registering a sentence in the database of the sentence prediction input method, the key sequence corresponding to the sentence and the key input sequence of each word constituting the sentence are key inputs corresponding to the initial consonant of each word. If a sentence is predicted even if it consists of , the key input corresponding to each word is reduced by more than half, making sentence prediction input more convenient. In other words, the key sequence corresponding to 'I go to school' is '22', which is the initial consonant of 'I', the first word, 'ㄴㄴ', and 'ㅎㄱㅇ', which is the initial consonant of the second word 'to school'. If you register '614', which corresponds to 'ㄱㄴㄷ', the initial consonant of the last word, 'I'm going', as a key sequence, you can predict and easily enter 'I go to school' even when you enter the initial consonant of each word. It will exist.

No sentence key sequence One I go to school. 21252-9-6111347-9-1162531 2 I go to school. 22-9-614-9-123

The sentence prediction input method of the present invention can register sentences containing mixed English and Korean in the database without distinguishing between Korean and English, even if English and Korean are assigned to an ambiguous key. As shown in Table 10, the English alphabet is also designated as an ambiguous key, so if you calculate and register the key sequence accordingly, you can freely register and extract sentences composed of English and Korean without having to separate English and Korean into separate keyboards. Even numbers, as well as English, can be assigned to ambiguous keys to enable mixed use of numbers, English, and Korean. The keyboard for this is shown in Figure 25. Unlike English and Korean, in the case of numbers, in order to specify only one number for one ambiguous key, the numbers '5, 6, 7, 8, 9' are entered with two keystrokes, and to display this, ' 5, 6, 7, 8, 9' are shown in pink. This means that '5' means '+5' (the pink square on the 'ㅁㅇ' key in Figure 25 means +5, and when combined with '1', it means '6', and when combined with '2' it means '7' .) is designated by pressing the 'ㅁㅇ' key and '0' is designated ('6', '7', '8', '9' is by pressing the 'ㅁㅇ' key corresponding to '+5'. Next, press the keys corresponding to '1', '2', '3', and '4' respectively) to ensure that all numbers are accurately distinguished. The reason for doing this is that in the case of letters, the words created by the combination of characters specified in each key are limited, but in the case of numbers, if two or more numbers are specified for an ambiguous key, numbers that can be combined into multiple ambiguous keys are pressed. Since the number increases in proportion to the number of keys, only one number should be assigned to the ambiguous key. Figure 25 shows an example of a sentence extracted through key input using an ambiguous keyboard in which numbers, English, and Korean are designated together. When the user presses and inputs the key corresponding to the key sequence '3912' corresponding to '2NE1' to search for a song by singer '2NE1', the song titles of '2NE1' are entered as shown in Figure 25. Sentence prediction input method It is extracted from the database and displayed in the prediction sentence display window. Accordingly, the configuration of the present invention provides a method for predictive input of English, Korean, and numbers on the same keyboard without mode conversion.

As shown in Example 34, the configuration of the present invention allows English to be input along with Korean, so that web addresses can be easily input without keyboard conversion for English input. In particular, by considering the '.' (period) symbol in the web address as a space symbol and treating it like a sentence, the web address can be registered in the database for the predictive input method as shown in Table 22, and the web address can be entered using the sentence predictive input method. There will be. Furthermore, as shown in Figure 26, when the cursor is placed in the web address input menu to input a web address, the database is set to search in an area that separates only web addresses for sentence extraction so that only the web address can be selected from the database of the sentence prediction input method. When configured, it increases the speed of search and enables efficient sentence prediction. Figure 27 shows a case where the searched sentence consists of only a web address when the cursor is located in the web address bar as shown in Figure 26. The purpose of extracting the desired sentences by subdividing the database configuration for the sentence prediction input method in this way is to minimize the number of sentences extracted from the sentence prediction input method so that the sentences to be input are extracted as prediction sentences.

web address key sequence first second third www.google.com 111 63361 739 www.bbc.com 111 997 739 www.youtube.com 111 2322291 739 www.daum.net 111 4129 912 www.naver.com 111 94711 739 www.cnn.com 111 799 739

In the sentence prediction input method, the order of listing the extracted sentences is prioritized according to the frequency of the predicted sentences, so that the most frequently used sentences are listed first and the user can select them. This is applied to a single word in the case of word prediction input, and in the present invention, in addition to the frequency of the sentence, high priority is given when the key sequence of each word constituting the sentence matches the input key sequence. In this way, cases where the key sequence matches are prioritized. For example, in the case of Figure 28, the currently input key sequence is '22'. And the Korean word corresponding to this is 'you', the number corresponding to it is '11', and the English equivalent is 'TT'. Therefore, the sentences containing these have a higher priority than the sentence 'November Illusion', which contains 'November', which does not match completely, and are therefore arranged at the front in the prediction sentence display window. This is summarized in Table 23. That is, according to Table 5, for sentences extracted from the sentence prediction input method database, the key sequence of each word constituting the sentence and the key sequence of the corresponding input key are compared, and the number of words with the same key sequence is 'n'. and at the same time calculates the number of words that make up the sentence, 'm', and gives high priority to sentences with large 'n'. If 'n' is the same, sentences with small 'm' have a higher priority than sentences with large 'n'. Arrange sentences by assigning priorities to select the most appropriate sentence. If 'n' and 'm' are the same, the priority can be determined according to the frequency of each sentence (the higher the frequency, the higher the priority) and arranged. In light of these standards, 'November Illusion' in Figure 28 has n = 0 and the remaining sentences have n = 1, so 'November Illusion' is arranged last in the prediction sentence display window.

Priority
output order detail One Calculate the number (n) of input key sequences that match the key sequence that makes up the sentence 2 Calculate the number (m) of words that make up a sentence 3 Sentences with large n have high priority. 4 For sentences where n is the same,
Sentences smaller than m have high priority.

Figure 29 shows an example of a Korean keyboard for the sentence prediction input method of the present invention. What makes this different from the Korean keyboard shown in Figure 28 is that the 'ㅂㅍ' key is placed at the bottom left, and among the vowels, 'ㅜ, ㅓ, ㅔ' are arranged in order from the left to the top, and on the right are 'ㅗ, ㅏ, ㅐ' is arranged in order from the top. And, 'ㅡ, ㅣ, ㅢ' are assigned to the upper center key. The background for this arrangement is to make it similar to the arrangement of Cheonjiin in the case of consonants so that users can easily familiarize themselves with the positions of consonants, and in the case of vowels, it is easy to input compound vowels (ㅝ, ㅞ, ㅘ, ㅙ). In order to do this, 'ㅜ, ㅓ, ㅔ' and ㅗ, ㅏ, ㅐ' are arranged in a straight line, and to make 'ㅚ, ㅟ' input convenient, 'ㅜ' and 'ㅗ' are arranged next to 'ㅣ'. will be.

Figure 30a shows a swipe method to input the word 'few' in the word prediction input method (in the input method with a touch sensor, rather than touching the keys to be input one by one, the key at the position where the finger first touches the touch sensor is used as the first key) It starts with a letter and is entered by successively passing the keys corresponding to the letter you want to input while your finger is in contact with the touch sensor. 'few' is entered as a finger trace. And in this swipe method, even when the same key must be entered more than once, the input is completed with a single touch (during the dragging process, the character passes the designated key only once). The keyboard shown in Figure 30a is the Google Pinyin keyboard, which allows you to input Chinese and English (word prediction input method is applied). To input 'few' in the swipe method, touch the 'def' key and pass the 'mno' key. When the 'wxyz' key is touched and the fingers move apart, 'few' is input as shown in FIG. 30A. However, the number of predicted words shown in the left predicted word display window 302 of FIG. 30A is 20. In fact, in the keyboard system of Figure 30a, the prediction word display window 302, which can show predicted words, can show 16 words, so the last line 303 is hidden by touching the arrow 304 next to the prediction word display window. It must be displayed at (302) and then selected. In other words, key input has become faster with the swipe method, but when the space for displaying the predicted word is limited, it takes a long time to display and select the predicted word on the display window 302, which reduces the speed of the entire input process. It is correct. Therefore, in order to increase the overall speed of input, the number of predicted words must be reduced, and to do this, the swipe method must be supplemented in two ways. The method can be summarized in two ways as shown in Table 24 below.

Method 1 Distinguish between cases where the key input is made only once and cases where the key input is made more than once for the key that the swipe trajectory passes by. Method 2 If Method 1 applies:
Arrange the keys on the keyboard so that the swipe straight path does not contain more than 3 keys.

Method 1 is a configuration that corresponds to the algorithm that applies the swipe input method. Therefore, this solution only needs to predict words corresponding to one key input sequence (key sequence) in which key input is made once or twice or more for the key that the trace passes after the swipe trace is formed. In contrast, Method 2 is not realized by an algorithm, but involves a change in the keyboard shape by bringing about a geometric change in the key arrangement.

Looking at method 1 in relation to the 'few' input, the predicted word for the swipe trajectory 301 in Figure 30a includes 'few', which requires entering the 'def' key twice, and 'ex', which requires entering only once. This shows that Method 1 was not applied. Accordingly, if Method 1 is to be applied, the swipe trajectory 301 of FIG. 30A must be modified. In general, in order to apply the method in Table 1, if the same key must be input more than once, a circular trajectory is drawn within the same key area and the key is repeatedly input, unlike a trajectory that simply passes by. Alternatively, if the finger stays on the key for a long time and passes briefly, a single key input is made, and if the finger stays on the key for a long time, the number of inputs is increased according to the stay time. In this way, even if the swipe trajectory has the same geometric shape, the number of key input repetitions is differentiated by varying the stay time or the shape of the key stay, thereby preventing multiple key input sequences from occurring in the same swipe trajectory, Figure 30a Because the predicted words 'few' and 'ex' are distinct due to the trajectory 301, the trajectory drawn to input 'few' (lengthen the time your finger stays on the 'def' key, or press the circle on the 'def' key) For drawing operations), words such as 'ex', 'DX', 'DW', and 'ez' are not extracted from the list of predicted words. In other words, Method 1 in Table 24 includes repeated input of keys that the swipe trajectory passes by, thereby reducing the number of predicted words that increase. In relation to method 2 in Table 24, we will examine the input of 'few' in Figure 30a. The swipe trajectory 301 for inputting 'few' in FIG. 30A draws a circular trajectory with the 'def' key to satisfy method 1 in Table 24, or extends the dwell time and then presses the 'def' key and 'wxyz'. It must be modified to draw a trajectory connecting the keys in a straight line. Even if the trajectory is modified in this way, the keyboard of Figure 30a cannot avoid the swipe trajectory of the 'mno' key located in the middle of these keys, so words containing 'm', 'n', and 'o' are excluded only through the word prediction algorithm. I can't do it. If the trajectory is changed to avoid passing the 'mno' key, the neighboring key 'jkl' will be passed, making it impossible to avoid the 'mno' key. Therefore, in order to implement method 2 of Table 24, the keys are arranged so that no other keys are located within the straight line connecting any two keys on the keyboard. To this end, Figure 30b schematically explains the process of deforming the keyboard of Figure 30a. (a) of Figure 30b explains the situation in which method 2 of Table 24 should be applied. That is, as shown in (a) of Figure 30b, the trajectory 305 connecting the keys located at the corners (keys 1, 3, 5, and 7) is the key located in the middle (keys 2, 4, 6, and 8). Method 2 in Table 24 can only be implemented by changing the position of the key located in the middle. And (B) in Figure 30b shows that the positions of keys 2, 4, 6, and 8 were changed to implement method 2 in Table 24. In summary, in (a) of Figure 30b, with the keys 1, 3, 5, and 7 located at the corners of the keyboard fixed, the swipe traces 305 connecting the keys located at these corners are 2, 4, 6, and 8. These keys 2, 4, 6, and 8 are moved so that they do not pass through the center of the key, and in the case of key 0, there is no suitable position to move to, so the position of key 0 is left empty. And if you move keys 2, 4, 6, and 8 like this, the keys will naturally be placed at the vertices of an octagon. The point to consider here is how far the keys 2, 4, 6, and 8 must move from their original positions. If (c) and (d) of Figure 30b are explained as an example, it may be good for the 4th key to move as far as possible from the trajectory 306 connecting the 3rd key and the 5th key, but as the distance becomes longer, While the 4th key may move away from the trajectory 306, a situation may arise where the 2nd, 3rd, and 4th keys are located on the same line, so as shown in Figure 30b (b), the 2nd, 4th, and 6th keys , It is most preferable that key number 8 is arranged at the vertex of a regular octagon. Specifically, the trajectories 306 and 307 shown in (c) and (d) of Figure 30b are distinguished by the swipe trajectory judgment algorithm. Just adjust the distance within the range. To summarize the above, in the case of a keyboard with 8 letter keys, when applying the swipe method in the word prediction input method, the central area of the octagon is emptied to reduce the number of predicted words, and 8 letter keys are arranged at the vertices to create two The keyboard is configured so that the straight line trajectory connecting the letter keys deviates from the center of the other letter keys to minimize the number of predicted words. In other words, it is a configuration that ensures that the trajectory connecting the two keys you want to input does not pass through the center of the key you do not want to input, so that only one key input sequence is performed for the swipe trajectory.

The keyboard shown in Figure 31a has seven letter keys arranged at the vertices of a heptagon based on the configuration presented in Example 38. (If the centers of neighboring letter keys in the keyboard shown in Figure 31a are connected, a heptagon is formed. ) The regular heptagonal keyboard shape was applied to use the key arrangement of the English qwerty keyboard. The character groups 'qwer', 'tyu', 'iop', 'asdf', 'ghjkl', 'zxc', and 'vbnm' are assigned to the seven keys. If you look at its characteristics, it is in the form of keys arranged in positions corresponding to each vertex of a regular heptagon. The advantage gained from the swipe input method by arranging the keys in this way is that the finger trace pointing toward the key to be input does not place other letter keys between the two keys. For example, the finger trace for entering the Korean word 'mosquito' using the swipe method will be explained by comparing that created on the keyboard shown in Figure 31b with that created on the general keyboard type (Figure 31c). Figure 31c shows a grid-type keyboard composed of 9 keys in the form of a general keypad. The traces (red traces) in Figures 31b and 31c show the finger traces for inputting 'mosquito'. In other words, starting from the key designated with 'ㅁ', passing through the keys designated 'ㅗ' and the keys designated 'ㄱ' in turn, and finally reaching the key designated 'ㅣ', moving the finger away from the keyboard to input in a swipe manner. The way is done. From these swipe trajectories, 'mosquito' can be predicted and entered. However, in the trajectory shown in Figure 31c, the same trajectory occurs even when the word 'grid' is entered in the swipe method in addition to 'mosquito'. In other words, in the process of moving to the key designated by 'ㅗ' and the key designated 'ㄱ', an unwanted key (the key designated 'ㄴ') is passed over, creating a situation where the unwanted key is pressed. In contrast, in the key arrangement of the keyboard shown in Figure 31b, the trajectory of moving from the key designated 'ㅗ' to the key designated 'ㄱ' does not pass through other key areas, so unlike Figure 31c, unwanted keys are not input. That is, in the key arrangement on the keyboard shown in Figure 31b, there are no other keys on the straight path connecting one key to the remaining six keys, so the swipe input method uses the key corresponding to the spelling of the word to be input. The goal is to get the same results as pressing them one by one. That is, in the case of the general grid keyboard shown in Figure 31c, when passing three keys located in a straight line, it is unclear whether the middle key must be input or not, so a key sequence that does not include input of the middle key and input of the middle key are used. Due to the key sequence containing , two key sequences occur. If the swipe trajectory passes through three keys in a straight line twice, a total of four key sequences are generated, two for each trajectory passing through the three keys, and only one key sequence is generated, as shown in Figure 31b. The number of words to be predicted increases four times compared to the keyboard shown in . Therefore, the selection process becomes difficult due to the increased number of predicted words. When this situation is applied to sentence prediction, assuming that there are three words that make up the sentence and that four key sequences occur for the swipe trajectory for each word, the number of key sequence combinations for sentence prediction reaches 64. Furthermore, if 4 sentences are predicted for each key sequence combination, the total number of predictable sentences becomes 265. It reaches the point where it is practically impossible to select the desired sentence from this large number of predicted sentences. In this regard, when entering an ambiguous key keyboard using the swipe method, it is important to arrange the keys so that unwanted keys are not located on the path connecting the key to be input. In this regard, as shown in Figure 31b, the keys are arranged at the vertices of the heptagon (connecting the centers of each key in Figure 31b creates a heptagon) or on each side of the heptagon, so that when entering in the swipe method, each key It is possible to prevent unwanted keys from being located on the shortest path connecting . The configuration of Example 38 and this embodiment are the same in that character keys are arranged at the vertices of a polygon, and the swipe trajectory connecting two vertices has only one key input sequence. However, in the heptagonal keyboard shape of this embodiment, the keys of the commonly used qwerty keyboard are grouped into 7 and assigned to 7 keys so that the user can easily recognize the key arrangement (array of letters) within the keyboard. Example 38 The octagonal shape of the keyboard has been transformed into a heptagonal keyboard.

In the swipe input method by placing keys at the vertices of the heptagon seen in Example 39, the unity of the key input sequence for the swipe trajectory (when there is only one key sequence corresponding to one swipe trajectory) In order to maintain this, it is necessary to ensure that the straight line path from each key to its neighboring keys, especially the second neighboring key, does not pass through the center of the immediately neighboring key. For example, the trace 307 shown in FIG. 32A passes through the center of key 4, so key input is made, and the trace 306 passes through an area deviating from the center of key 4, so key input is not made. In other words, it is determined whether the trajectory passed through key 4 according to the degree of bending of the trajectory in the key 4 area. The content of this trajectory judgment standard must be adjusted according to the pointing device that specifies the point where the touch is made on the keyboard. For example, if the pointing device that touches the keyboard can accurately know the touch point, such as a stylus pen, the user can accurately distinguish and follow the trajectories 306 and 307 of FIG. 32A. However, in the case of touching using a finger, the actual finger touches It is impossible to know whether the point is the center of the key or not. Therefore, the user intends to follow a trajectory corresponding to 306, but may actually move along trajectories 308, 309, and 310. Therefore, the standard for judging the trajectory considering the various trajectories (306, 307, 308, and 309) shown in Figure 32a is whether the trajectory passes the key, as well as a straight trajectory such as 306 and a curved trajectory such as 307. A standard is established by distinguishing differences in trajectories. In other words, whether or not key 4 is input is determined depending on the geometric shape (degree of bending) of the trace passing through key 4, so even if the trace passes through the center area of the key as shown in Figure 32a (e), key input is made. There are times when you don't lose. Therefore, if the standard created in this way does not match the trajectory created by the user's finger or pointing device, users must go through a process of adapting to this standard through trial and error or change the trajectory judgment standard to suit the user's habits. Just do it. In other words, the swipe input method has a higher probability of causing an error compared to the touch input method where each key is input. In order to reduce this error rate, we would like to consider reducing the size of each key placed at the vertex of the polygon. One of these methods is to reduce the size of the key placed at the vertex of a regular polygon, as shown in Figure 32b. The regular polygons shown in FIG. 32b are shapes inscribed in a circle of the same size as shown in (a) of FIG. 32b. In addition, the size of the key placed at the vertex of each shape is adjusted so that the trajectory connecting two keys does not pass through other keys, and the key in Figure 32b has the maximum size that a key can have when placed at the vertex of a polygon. It shows the size. That is, as shown in (c) to (b) of FIG. 32b, the straight line trajectories (324, 325, 326, 327) connecting key number 1 of each regular polygon to key number 3, which is the second closest, are when the user moves from key number 1 to key number 3. This becomes the outer boundary of the area 320 of the trajectory moving with the number key. Accordingly, in the case of key number 2, it comes into contact with the boundary lines 324, 325, 326, and 327 of the trajectory area 320 connecting key number 1 and key number 3. The structure of this embodiment is to adjust the size of the keys arranged at the vertices of the regular polygon so that the trajectory connecting two keys does not pass by other keys. Therefore, in the keyboard form of Figure 32b, the criterion for judging the swipe trajectory is set to include "all keys that the trace passes through in the key input sequence (key sequence)," so if the user allows the swipe trace to pass only the desired keys, the swipe trajectory can be adjusted. You have the advantage of not having to worry about it. That is, in the key arrangement shown in Figure 32a, the user cannot accurately determine the trajectory judgment criteria that distinguishes between trajectory 306 (no key input of key 4) and trajectory 307 (execution of key input of key 4), so draw the trajectory and then input it. While it is necessary to check whether the input has been entered correctly by looking at the result, in the key arrangement of FIG. 32b, the key area that is passed can be visually checked, so the user can visually check whether the trajectory meets the judgment criteria, and the trajectory of FIG. 32a Since the confirmation process based on judgment criteria can be omitted, quick and convenient input is provided.

As shown in Figure 32b, in order to satisfy the condition that the straight line trajectory connecting two keys does not pass through the area of another key, the maximum size that each key can have becomes smaller as the number of vertices of a regular polygon increases. . Furthermore, the key shown in FIG. 32b is the maximum size that can not pass another key, which is a trajectory connecting two keys, so when the size of the key becomes smaller, the trajectory connecting keys 1 to 3 in FIG. 32b becomes key 2. The probability of passing by decreases. On the other hand, if the user wants to increase the size of the key while maintaining the distance between the keys to facilitate key selection, the number of vertices of the polygon must be reduced as shown in FIG. 32b. In other words, the number of keys must be reduced. Conversely, if you want to increase the number of keys, the number of vertices of a regular polygon increases, and accordingly, the size of the keys placed at the vertices of the regular polygon cannot help but decrease. In other words, the number of keys and the size of the keys are inversely proportional. If you want to increase the size of the key to facilitate key selection, the number of character keys placed in the polygon must be reduced. In this case, as a result, the number of characters to be assigned to each key in the predictive input method is It leads to many disadvantages. In other words, as the number of characters assigned to one key increases, the number of words predicted from the same key input also increases, which increases the size of the key and makes key selection easier, but the number of predicted words increases and the number of words increases. This has the disadvantage of making it difficult to select the desired word from among the predicted words. In order to compensate for this shortcoming, a method of maintaining the unity of the key sequence for the siwpe trajectory without reducing the number of keys by increasing the key size is shown in FIG. 33. In summary, if you increase the size of the key placed in the same regular polygon, a part of the siwpe trajectory will overlap with an unwanted key. By disabling this overlapping part, the unity of the key sequence for the swipe trace will be maintained even if the key size increases. This is the way to do it. Explaining in detail the content of FIG. 33, in FIG. 33a, a key is placed at the vertex of a regular heptagon, and the unity of the key sequence is maintained for all straight trajectories connecting keys 1 and 3. Figure 33b shows the diameter of the key being doubled. If the diameter of the key is doubled, the area through which the straight line connecting keys 1 and 3 (the area between 333 and 335) passes will overlap about half of key 2. If this overlap is possible, both the key input sequence for the trajectory connecting keys 1 and 3 are '1'-'3' and '1'-'2'-'3'. As this breaks down the unity of the key sequence for the swipe trajectory, the area through which the straight line passes is reduced by moving trajectory 333 to 334 and overlapping this virtual area (the area between 334 and 335) with key number 2. The area is considered a non-overlapping area. If the non-overlapping area is defined in this way, if the contact point of key 1 remains in the area (336) before the area is expanded, the possibility that the movement trajectory to key 3 overlaps with key 2 is low, so key input for the swipe trajectory The key sequence is determined as '1'-'3'. In general, when touching a touch sensor with a finger, the finger is approximately located close to the center of the key, so the actual user's swipe process is similar to the process that occurs in the key arrangement as shown in Figure 33a. Therefore, as shown in Figure 33c, the area obtained by applying the reduced area (area between 334 and 335) of the connection trace of keys 1 and 3 to keys 2 through 7 in order (key 2 and If you combine the area between the 4th key, the 3rd key and the 5th key, the 4th key and the 6th key, the 5th key and the 7th key, the 6th key and the 1st key, and the 7th key and the 1st key, A regular heptagonal area 337, which is a virtual area, is formed. In other words, this virtual area is defined as a non-overlapping area so that if this area is passed during the swipe process, the finger contact point (if the contact point corresponds to a mouse cursor, the cursor position point) passes the key, it is not recognized as a key input. This is a design of an embodiment. Based on this, in order to increase user convenience, a method may be provided to adjust the distance from the virtual trace line 334, which defines the size of the key and the non-overlapping area, from the trace line 333 as desired by the user. there is. Additionally, it is possible to provide a method of defining the swipe trajectory 334 as an arc rather than a straight line and applying it to the user's finger movements (or mouse cursor movements). By setting the non-overlapping area provided in this embodiment, it is possible to increase the size of the key without damaging the unity of the swipe trajectory, thereby promoting ease of key selection and convenience of swipe operation. Furthermore, as shown in Figure 34, various geometric shapes other than the shape of the key can be applied, and as shown in Figure 34c, additional keys are placed within the non-contact virtual area 337 to add numbers 1 to 7 such as symbols and spaces. You can add non-character input and various other functions that the keys are not responsible for. In addition, these keys (keys 'a' to 'sa') located within the non-contact virtual area 337 are located within the non-contact virtual area, so even if the swipe trajectory for the word prediction input method passes these keys, numbers 1 to 7 No input is made due to these keys in the swipe operation initiated from the keys. However, if the contact starts with these keys and the contact ends with these keys, the function assigned to these keys is performed, and if the contact ends with these keys, the function assigned to these keys is performed. When a drag (swipe) operation is continued, additional functions can be performed according to the trajectory to increase the diversity and efficiency of input.

In the word prediction input method, when key input is performed using the swipe method, if the same key must be entered more than twice, the word can be predicted with just one swipe. In other words, the swipe trajectory shown in Figure 35a is a trajectory for inputting 'need'. Start touching on key 5, where 'n' is designated, and move to key 1 to input 'e' with your finger on the touch sensor (if you pass key 1, you will not only find words containing only one 'e', , words containing 'e' more than once are also extracted from the database), then move to the 7th key where 'd' is designated and then lift your finger from the touch keyboard to extract 'need' (not only 'need' but also 'e' 'Ned', which contains only one, is also extracted) and entered. In this case, the key input sequence is interpreted as '5'-'1'-'7' and '5'-'1'-'1'-'7'. However, in the word prediction input method, when input is made using the swipe method, as shown in Figure 35a, when a key is passed along a swipe trajectory and a word containing only one character specified in the key is predicted, there are two 'e' as in 'need'. Since it is included, you must pass key number 1 twice, but on the general grid keyboard shown in Figure 31a, there is no way to pass it twice, so you can pass it twice by lengthening the time you stay on that key or making a circular motion on that key. It brings about the same effect as . On the other hand, in the keyboard of the present invention shown in Figure 35, if a key is dragged to an area outside the area (an area including a non-contact virtual area) and the key is returned, the key area passed twice results in passing the key area (5). The 'need' is predicted in the order of -(1)-(1)-(7). On the other hand, in the case of Figure 35a, the key area that the finger passes through is (5)-(1)-(7), so 'need' composed of 4 letters cannot be predicted, and instead 'Ned' composed of 3 letters is predicted. It is predicted. Figure 35(c) shows the trajectory of entering 'pioneer'. This is because the trace in Figure 35(b) moves to a non-overlapping area to enter 'e' in 'need' twice and returns to the key number 1, while the trace in Figure 35(c) moves to the non-overlapping area to enter 'e' in 'need' twice. To input 'p', 'i', and 'o', move to the outside of the non-overlapping area (outside the heptagon connecting the letter keys), then return to key 3 and press the same key 3 times. The idea is to recognize it as a process. Likewise, to input 'eer' in 'pioneer', after passing key 5 for inputting 'n', it moves to key 1, moves to the area outside the non-overlapping area, escapes key 1, and returns to key 1. When the fingers move apart, it is recognized as the process of entering the number 1 key corresponding to 'eer' three times, and the trajectory in 35(c) becomes the swipe trajectory for entering 'pioneer'. To summarize Figures 35b and 35c, while touching one key, you move out of the non-overlapping area. When you return to this key, you enter the key once more and move out of the way to the outer area of the polygon corresponding to the non-overlapping area. The idea is that when you come back, you will have to press the key twice more. In this way, the convenience of key input can be increased by distinguishing the inner and outer areas of the polygon connecting the character keys and designating them as the area through which the swipe trajectory passes. In addition, by matching the number of keys passing through the swipe trajectory and the number of letters constituting the word, the number of predicted words is reduced, making it easier to select the predicted word. Furthermore, in the case of sentence prediction input, if the swipe trajectory allows duplicate input of keys that pass by, the key input sequence for each word constituting the sentence becomes multiple, and as seen in Example 38, the predicted sentence The number is so large that it is difficult to choose. Therefore, the configuration of Example 38 and this embodiment provides a keyboard with increased convenience and accuracy that satisfies key sequence (key input sequence) unity for the swipe trajectory. To summarize the above, Embodiment 38 provides a configuration to prevent unwanted key input, and this embodiment distinguishes between the case of entering the same key once and the case of entering the same key multiple times, and only one key input is required for the swipe trajectory. Provides a method to minimize the number of predicted words by providing an order.

Unlike languages such as Korean and English that include spaces between words that make up sentences, Chinese and Japanese do not include spaces in their sentences. The criteria in Table 5 are not applicable to these languages. Then, in order to apply the standards in Table 5 to non-space languages that do not contain these spaces, the elements that make up the sentence are subdivided and then spaces are inserted between these subdivided components. Simply put, even non-space languages use spaces to separate search keywords when doing Internet searches. These keywords used in search are used as elements to separate sentences to form sentences that include spaces. Chinese can be compared to English in its similarities, and Japanese can be compared to Hangul in its similarities. In Japanese, like in Korean, the syntactic function within a sentence is determined by combining a noun with a particle, and in the case of a verb, the tense is revealed through a change in the ending, so a noun phrase combined with a noun and a particle is an element that makes up a sentence. In the case of verbs, verbs combining a stem and an ending are treated as elements that make up a sentence. In other words, if you look at the Japanese sentence 'I go to school' below, which means 'I go to school', the Korean language has 'I' corresponding to the subject, 'To school' corresponding to the adverb clause, and 'To school' corresponding to the verb. 'I'm going' is separate, but in Japanese, '私(は)' corresponds to 'I', '學校に' corresponds to 'to school', and '行きます' corresponds to 'going'. ' is displayed attached to each other instead of being separated. Therefore, in relation to the structure of Table 5, Japanese sentences can be displayed as '私は 學校に 行きます。' by applying spacing to have the same grammatical structure as Korean, and sentence prediction input according to the standards in Table 5 This method allows you to input Japanese. In other words, even in the case of Japanese without spaces, the verb part is separated from the verb part, and the verb part is made up of a combination of a verb and a particle, and the verb part is made up of a verb and a ending. In addition, the adverbial part and the object part are separated from the noun part and the verb part, so that the sentence prediction input method can be applied according to the standards in Table 5. Therefore, the search method according to the sentence prediction input method shown in Examples 29 and 30 can be applied to Chinese and Japanese in the same way as to sentences containing space characters such as Korean and English.

I go to school.

Watashiwagakkoniikimasu

(Watashiha // gakkoni // ikimasu)

In Example 26, a database for predicting noun parts and verb parts of Hangul was constructed as a separate database of nouns and particles, and stems and endings of verbs, and extracts nouns and particles from these separate databases to predict noun parts and predict verb parts. It was shown that verb parts can be predicted by extracting stems and endings. Through this, it was possible to prevent the body part and verb part database from growing large. This database construction method can also be applied to compound noun groups. A compound noun is when two or more nouns are combined to create a new noun. If all compound nouns are built into a database, the size will be enormous, so as a way to simplify this, a method is provided to combine the nouns that form a compound noun. do.

The specific method is that when the 'space' key is pressed twice in succession, the word predicted by the key input made after the 'space' key is pressed is combined with the word predicted by the key input made before the 'space' key to form one (compound ) word so that it is listed in the predicted word list. In other words, to enter the phrase 'Gyeonggi-do Rehabilitation Engineering Service Research and Support Center', enter the keys corresponding to 'Gyeonggi-do', 'Rehabilitation', 'Engineering', 'Service', 'Research', 'Support', and 'Center'. When the key corresponding to each word is entered, the 'space' key is entered twice to predict and extract the following words as a single connected phrase without separating them with spaces. This method can also be applied when entering a name. In other words, if the proper noun 'Gang Chan-gam' is not registered in the database, dividing each syllable into 'gang', 'chan', and 'gam' by entering the 'space' key twice, 'Gang' Since the words (vocabularies) ', 'chan', and 'gam' are already registered in the database, these vocabularies can be combined to make it easy to enter the name 'Gang Chan-gam'.

Therefore, in this Example 44, in the case of Hangul, compound nouns and proper nouns (last name + first name) are written together unlike in English, so only the words that make up each can be registered in the database and the compound nouns can be extracted by combining these words. It provides a method. This allows the size of the database to be reduced and further provides a way for users to freely enter compound nouns that are not registered in the desired database. For user convenience, the method of entering 'space' twice can be implemented using a separate key. Furthermore, the functions of pressing 'space' once and twice can be interchanged to enable double pressing 'space'. You can perform the 'space' input function, and press 'space' once to perform the function for combining nouns.

As seen above, the sentence prediction input method of the present invention can improve the convenience and speed of input even with a small number of keys, so it can demonstrate its effectiveness in mobile devices such as smartwatches, and furthermore, it can be used to increase the number of keys such as remote controls. It will be applied to limited input devices and will serve as an opportunity to further expand the usability of these devices by overcoming the situation where remote controls could not be used for text communication until now. In the future, remote control-type input devices that will be used in VR (virtual reality) and AR (augmented reality) will provide efficient text input using 8-way keys. For example, in the case of recently popular VR remote controls, joysticks are used to adjust direction, and this embodiment provides a method of using these joysticks to input text, providing a method to easily input text in VR.

Figure 38 shows the keyboard shown in Figure 31c transformed into a circular shape to enable key input using a joystick. 38(a) and 38(b) are pointers that move within the keyboard according to the movement of the joystick and are in the form of a sphere. These pointers are moved with the left and right joysticks, and when these pointers return to their original central position after contacting a key, an input corresponding to the key that these pointers last touched is made. And these two joysticks move the pointers (spheres) of 39(a) and 39(b), respectively, doubling the input speed.

In fact, when inputting with the keyboard shown in Figure 38, when inputting using only one joystick, the input can be achieved at about 100 to 120 strokes per minute, whereas when inputting by operating two joysticks with both hands, the input can be achieved at about 200 to 250 strokes per minute. Input speed is doubled.

Looking at this from a UI perspective, a computer with a GUI structure uses only one cursor because there can only be one focus that the user can visually track. Assuming that there are two pointers (cursors) on the screen and the user manipulates the two pointers with the mouse, in order to manipulate the pointers (cursors), the user must visually focus on the pointers and move the pointers. You operate the mouse while checking . Therefore, since the user cannot operate two pointers (cursors) at the same time, even if there are two pointers on the screen, the user cannot visually focus both pointers at the same time, so the number of pointers cannot be adjusted at the same time. Even if it increases, there is no advantage. Sometimes, in special cases, a multi-pointer (cursor) system that provides multiple pointers to one window is used. This is not for single-use computers, but when several people work together by sharing one screen, each person has a separate window. The purpose is to provide a pointer (cursor) so that each individual can work independently with one pointer (cursor). Therefore, if you want to operate two pointers like the keyboard in Figure 38, it is not possible in a mouse system that controls the pointers through visual focus confirmation. The multi-pointer method provided in this embodiment uses a method of controlling the pointer by checking focus through the sense of touch, rather than a method of controlling the pointer using visual focus. In other words, instead of a mouse based on visual focus, a joystick is used as a pointer control device. The difference between a mouse and a joystick as a pointer control device is that in the case of a mouse, the position change is determined by the displacement of the pointer (mathematically, delta-x, delta-y), whereas in the case of a joystick, the position value of the joystick is converted into the coordinates of the pointer. The point is to use it. To explain mathematically, in the case of a mouse, in order to calculate the new position of the pointer, the displacement (delta-x, delta-y), which is the difference between the current coordinates of the mouse and the coordinates of the moved position, is used to move the pointer. Against this background, when the mouse is lifted from the mouse pad and moved to a new location, the mouse is separated from the mouse pad, and the displacement due to the movement of the mouse is not applied to calculating the pointer position, so the pointer remains stationary and does not move. Even if the mouse position changes, the pointer position does not change, so there is no 1:1 correspondence between the mouse position and the pointer position. This method is called the relative coordinate method, and as a result, pointer control based on the relative coordinate system cannot provide eyes-free functionality. Unlike the relative coordinate mouse, the joystick uses absolute coordinates to determine the pointer's position based on the joystick's position. To explain using the pointer (sphere) shown on the keyboard in FIG. 38, if the joystick is at the center of the device, the pointer is also located at the center of the keyboard shown in FIG. 38. To compare this to a 4-way key, when you move the joystick north, the pointer also moves north, and when your finger lets go of the joystick, it returns to the center position due to the built-in spring, and the pointer also returns to the center of the keyboard. In other words, it is an absolute coordinate system in which the position of the joystick and the position of the pointer correspond 1:1. However, in the case of a mouse, if you lift the mouse from the mouse pad and move it to a different position on the mouse pad, the pointer does not move, so the position of the mouse on the mouse pad does not correspond 1:1 to the position of the pointer. Therefore, when moving a pointer using a mouse, the location cannot be confirmed unless the pointer is always visually focused and tracked. In this respect, mouse pointer control cannot be eyes-free. On the other hand, the pointer control of the joystick can be performed eyes-free. This means that the finger recognizes the position of the joystick, and the position of the pointer corresponds 1:1 to the position of the joystick, so the user can determine the position of the joystick just by touch and control it accordingly. The position of the pointer is also recognized. Therefore, unlike when using a mouse, only one pointer can focus, when using a joystick, the left and right fingers can simultaneously focus on the positions of the two pointers on the keyboard shown in Figure 38 by tactile sense. This advantage allows the user to simultaneously control the pointer (sphere) on the screen. This can also be compared to being able to operate a keyboard with ten fingers. When inputting on a computer keyboard, the user confirms the position of the ten fingers through contact, so it is possible to recognize which key the finger movement can lead to input without looking at the position of the fingers. Therefore, when you move the joystick, your finger can recognize the position of the joystick in eight directions, including east, west, north, south, and furthermore, southeast, southwest, northeast, and northwest, so even if you do not see the movement of the pointer, you can know which direction the pointer moved and which key is pressed. You can check with your sense of touch whether the input will be entered. However, the additional display shown in Figures 39 and 40 (changing the color of the key where the pointer is located) merely serves the function of confirming the movement of the finger. Against this background, the configuration of this embodiment provides a way to overcome the slow input speed when using only one joystick by configuring two pointers to be manipulated with two joysticks. This brings input efficiency comparable to that of a typical qwerty keyboard, reaching 200 to 300 strokes per minute, allowing input on VR devices to be performed like smartphones.

The keyboard shown in Figure 38 consists of only eight keys in the circumferential direction, and is characterized in that all keys can be selected by the circumferential movement of the joystick. Due to these features, the joystick cannot be applied to keyboards with more than 10 character keys, such as the qwerty keyboard, but by using two pointers along with the sentence prediction input method, it is easy to input characters using the joystick, which is a general direction control device for VR devices. It can be done quickly.

In conclusion, the configuration of this embodiment is to provide an eyes-free input device that doubles input efficiency by utilizing two joysticks.

Figure 39 is a diagram showing that the color of the key changes when the left and right pointers (sphere) are moved and positioned on the key on the keyboard according to the configuration of Example 45. This is a configuration that allows the user to easily recognize the location of the pointer by differentiating the color of the key where the left pointer is located from the color of the key where the right pointer is located. Furthermore, if the left and right pointers overlap, the color of the key where the pointer is located is displayed in a different color from the color (blue and pink) that appears when each pointer is located (in the case of Figure 40, when the left and right pointers are located at the same time, the color of the key is displayed) (color displayed in yellow) The configuration of this embodiment 46 is such that the user can recognize that the pointer is located on the same key, making it easier to visually determine the location of the pointer in any case.

This embodiment provides a configuration that provides the same eyes-free input device by using a touch pad (or a touch device that can control x and y directions, such as a touch screen) instead of the joystick described in embodiment 45. As already explained in Example 45, in order to become an eyes-free input device, it must be configured to recognize the position of the pointer through the sense of touch, and the position of the input device and the pointer must be structured in a 1:1 correspondence. Therefore, the touch device used in this embodiment must control the pointer based on the absolute coordinate system. In general, touch devices can be divided into touchpads used in laptops and touchscreens used in tablets. The difference between the two is that the touchpad controls the pointer (cursor) based on a relative coordinate system like a mouse, whereas the touchscreen controls the pointer (cursor) based on an absolute coordinate system. Controls the pointer (cursor) based on . However, the touch screen is a state in which the touch device is integrated into the screen and does not require a separate pointer, so you can think of the position of the finger as the position of the pointer.

Therefore, if you create embossing or convexities on a touch device (touch pad, touch screen, etc.) that provides You can move your finger (and consequently the pointer) in the desired direction or location using only your sense of touch. By using two touch devices with this configuration, the efficiency of input can be doubled, as in the case of a joystick.

Figure 41 shows the controller of the VR device. 41(b) is a touch pad for direction control, and the user's thumb touches the touch pad area to control the direction. In other words, unlike the touchpad of a typical laptop that is based on a relative coordinate system, this is a touchpad that allows direction control based on an absolute coordinate system. The size of this touchpad is such that the entire touchpad can be touched with the movement of the (thumb) finger joint, and the shape is circular, so when you touch the edge of the touchpad, the shape of the part your finger touches changes according to the eight directions of north, south, east, west, and south. Therefore, you can recognize which part of the edge of the touchpad your finger is touching. In this regard, the touchpad in Figure 41(b), even without separate embossing or irregularities, is the center of the touchpad in the situation where the user is currently touching the touchpad, or is it located at the edge in all four directions, such as east, west, north, south, and southeast, southwest, northwest, and northeast. It is possible to determine which direction the location is. A touch device with this structure can be applied as a pointing device with an eyes-free function that can operate the pointer of the keyboard shown in FIG. 38. Accordingly, the VR controller in Figure 41 shows an example of an eyes-free function that can be performed without separate embossing and unevenness by changing the touchpad structure itself. Nevertheless, embossing and irregularities can be provided on the surface of the touch device so that the user can more easily recognize the position of the finger within the touchpad.

In particular, pointer control using a touch device shown in FIG. 41 increases the input speed compared to a joystick. The reason is that in the case of a joystick, the joystick must always return to the center and then move to the desired key position, whereas in a touch device, there is no need to return to the center of the touchpad, but simply remove your finger from the desired key position (or the touch device has a press function). If you press the touch device and release the pressed state while the touch is maintained, key input can be made. Moreover, if the joystick also has a press function, you can move the joystick to the center when performing key input with the press function. (You can move the joystick to the desired key position without doing so.) Since the key input is made, the finger can be moved in a straight line, increasing the input speed. The most comparable example is when entering the 'qwer' key on the keyboard in Figure 38 and then entering the 'tyu' key right next to it. In the case of a touch device, you can release your finger from the 'qwer' key and move it right to the side, but in the case of a joystick, Since the pointer must be moved to the center while placed on the 'qwer' key and then moved to the 'tyu' key, the input speed is inevitably relatively slow. In fact, when inputting with a touch device based on the absolute coordinate system, 200 to 250 strokes per minute are possible with just one hand, which is better than a joystick in terms of input efficiency. However, in the case of Oculus, the world's most popular VR device, the joystick is used as a direction control device, so other functions must be considered not only as a text input device, so the text input method of the present invention using two pointers accordingly, whether it is a joystick or a touch device. By applying , you can achieve an input speed of 200 strokes per minute without feeling frustrated with the input speed. Moreover, when using the touch device of FIG. 41 as a pointing device, if two touch devices are used, an input speed of 200 to 250 strokes per minute is expected to be further improved than when using the touch device alone. Therefore, the text input method of the present invention provides an input system with an eyes-free function suitable for the controller in VR, AR, and XR devices that require eyes-free because the input device cannot be seen, enabling easy and fast text input.

Recently, visual tracking systems have been highlighted as a way to increase the efficiency of VR devices. VR devices have limitations in reducing the size of the device in order to process graphic information displayed on the screen. In this regard, the eye tracking system provides a method of determining the location of the user's visual focus on the screen, increasing the resolution around it, and lowering the resolution of the remaining areas. By doing this, the graphic information processing amount of the VR device is drastically reduced compared to the case where the resolution of the entire screen is increased, allowing the device to be smaller in size and increasing efficiency by reducing energy consumption.

Therefore, the eye tracking system also provides a method for controlling VR devices by providing a mouse system that can control the cursor. In this regard, the eye tracking system is used as a text input method using a virtual keyboard by providing a cursor control method called 'eye mouse' to people who cannot move their bodies freely. Figure 43 shows a virtual keyboard for an eye mouse. The method of use is to place the cursor on the key to be input and then have the key input after a certain period of time. In the virtual keyboard of Figure 43, the cursor is currently placed on key 'O' (43a) and is displayed as a red circle piece (pie) indicating the elapsed time. When pi, which represents the passage of time, forms a complete circle, a key input is made, producing the same effect as pressing a mouse key.

However, the general qwerty keyboard-shaped virtual keyboard of Figure 43, which is operated with an eye mouse, has the disadvantage that key input is made slowly, as it requires an elapsed time of at least 0.5 seconds. Moreover, if the cursor stays in one place for a long time, key input may be made unintentionally by the user. To prevent key input from being made, keep moving the gaze so that it does not rest on a single key, or keep the gaze focused on a specific location other than the key and move the eyes. It also causes discomfort by not being able to move. In other words, you experience the inconvenience of not being able to input keys quickly and not being able to move your eyes freely.

The virtual keyboard that provides a way to overcome these shortcomings and inconveniences is the keyboard of this embodiment shown in FIG. 42. In reality, the shape and arrangement of the keyboard and keys seen in Example 45 (FIG. 38) are the same, and the only difference between the keyboard of FIG. 38 and the keyboard of FIG. 42 is that the keyboard shown in FIG. 42 has a border 42c outside the key arrangement. is given. As explained in the paragraph above, compared to the general keyboard (Figure 43) that uses an eye mouse, key input in the keyboard shown in Figure 42 is performed by moving the cursor (42a - shown in the shape of a ball) on the center key marked in red to enter the desired input. If you pass the key and return to the center key, it will be entered. If the cursor goes beyond the outer boundary of the key (42c), the key input is invalid, and in order to input the key, the cursor must return to the center key, move to the key to be input, and return to the center key. In other words, if the cursor goes beyond the area outside the key, it is unrelated to the input, so the gaze does not have to stay at the designated location, and even if it moves freely, it has nothing to do with the input, so you can move your gaze as you like. Moreover, even if the cursor located on the outside of the key passes through the key in the process of returning to the center key, no input is made for the key passed at this time, so the user starts from the center key and returns to the center key without leaving the outside of the key. You only need to check the keys that the mouse uses. Moreover, unlike an eye mouse, there is no need to maintain a set time for the cursor to stay on a key when it passes by, which has the effect of increasing the input speed accordingly.

As seen above, the 45th embodiment and the present embodiment are the same in that the cursor starts from the center key, passes the key to which the character is assigned, and when it returns to the center key, input is made for the key that was passed. However, in Example 45, key input is made when the cursor returns to the center key, which means that key input is made when the joystick returns to its original stationary state. In contrast, in this embodiment, a key input is made when the cursor enters the area of the center key 42c. Therefore, when operating the cursor with an eye mouse, unlike a joystick, the mouse cursor cannot be automatically positioned at the center of the center key. Therefore, the starting point of key input does not need to be located at the center of the center key, but instead is located at the center of the center key. When placed within the area, key input begins.

In the case of the keyboard shown in Figure 38, since the cursor is manipulated with a joystick, the size of the center key or its function is simply to show the position of the cursor and does not actually perform other functions (such as functions other than character keys, such as menu expansion). It was placed for. In other words, when key input is made using a joystick, it is mechanically determined that the joystick is located at its original center, so even if the center key does not exist, there is no problem with key input. On the other hand, in the case of an eye mouse operated by eye tracking, if there is no center key, the starting point area cannot be specified, so it is absolutely necessary, and its size must be made as large as possible. For this reason, the core configuration of this embodiment is to arrange the character keys around the central key and to press the character key (in the case of an eye mouse, set to the time the cursor stays on the character key) so that the cursor starts from the central key to indicate the desired input. When you pass through the keys and return to the central key, the input is made. For this purpose, the configuration of this embodiment is to designate a starting area (center key area). By doing this, eye mouse users can freely move their gaze on the keyboard, and if they intend to input text, they can start by focusing their gaze on the center key. On the other hand, in the case of the general keyboard shown in Figure 43, there is the inconvenience of having to perform a 'clicking action' (blinking eyes) or 'staying on the key for a certain period of time' action to distinguish the key to be input from other keys, so 'character The core configuration of this embodiment is to designate an 'input start area' so that key input can be made only by moving the cursor, and it is desirable to place the 'character input start area' at the center of the key arrangement. The reason is that the same distance can be maintained for all keys, and there is no need to worry about passing other keys when returning to the 'character input start area'. Nevertheless, in some cases, it may take the form of a modified arrangement. In Figure 42, in addition to the character keys, function keys ('delete', 'menu' keys, etc.) are located on the outside of the character key array. In the case of these function keys, when the cursor starts from the center key and returns to the center key after passing through these function keys, In this case, the character keys passed after the cursor passes through the function keys are ignored. In other words, it is possible to have a concentric key arrangement in which keys farther away from the center key have input priority.

In the case of the sentence prediction input method of the present invention, when a key input is made, the predicted sentence and predicted word as shown in FIG. 44 are displayed in the predicted sentence list window 44a and the predicted word list window 44b, respectively. Predicted words are also displayed in each key area, providing an easy way to select them. Depending on the device that controls the keyboard shown in FIG. 44, the method of selecting predicted words and predicted sentences may vary. For example, if the keyboard in FIG. 44 is operated in conjunction with a joystick, the movement of the cursor 44f does not deviate from the key outer boundary 44d. In contrast to the joystick, if the keyboard in Figure 44 is operated in conjunction with the mouse, the cursor 44f has no restrictions on its movement, so select the prediction word or sentence you want to enter in the list window (click the area where the prediction word or sentence is displayed). You can input it like this: In fact, the keyboard in Figure 44 is used in conjunction with a remote control, so the remote control is equipped with a gyro sensor to provide an air-mouse function that allows the cursor to move according to the movement of the remote control, or a touchpad-like keyboard. If a pointing device is installed, move the cursor to the prediction word (sentence) list window and select (click) the desired prediction word (sentence) and enter it. However, if the keyboard in Figure 44 is operated in conjunction with a joystick and the movement of the cursor does not exceed the outer boundary of the key, selection input by mouse click is not possible, so selection input is made using a separate function key or the button function of the joystick. can do. In the case of a joystick where the movement of the cursor is limited to the outer boundary of the key, and for devices such as an eye mouse where mouse clicks are not free, a method of selecting a predicted word (sentence) is provided by displaying the predicted word (sentence) in the key area. The content is the configuration of this embodiment.

[1] Prediction word selection – joystick

Predicted words are displayed on each letter key in addition to the word prediction list window 44b, allowing for easy selection and input. When key input is made using a joystick, the key input is performed by moving the cursor to the key position (moving the joystick from the center position to the key position) and then returning to the center key area (when the joystick returns to the original center position). It comes true. If the joystick has a push function, if you move the joystick to the position of the key and press it, the predicted word displayed on the key is selected instead of key input. If the joystick does not have a press function, designate a separate press switch, move the joystick to the position of the key while pressing the push switch, and when it returns to the original position, select the predicted word specified at the position of the key, and the joystick will function as a press function. Even if there is no button, the push switch is responsible for the pressing function of the joystick.

[2] Prediction word selection - eye mouse

In Example 48, the keyboard configuration for an eye mouse was explained. In this case, compared to a general mouse, the eye mouse's clicking action is generally performed by blinking, so the pressing function of the joystick is replaced by blinking to take charge of the selection function of the predicted word. Just do it. However, if eye blinking is responsible for the clicking action, there is a disadvantage that it may cause confusion if the eye blinks due to a physiological phenomenon even if the clicking action is not intended. Therefore, in order to exclude the eye blinking movement for mouse clicking and select predicted words (sentences), the key outer area is subdivided into a key input outer boundary (44d) and a word selection outer region (44e). By dividing the key outer area in this way, when the cursor moves beyond the key input outer border 44d, moves to the word selection outer area 44f, and then returns to the central key area, the prediction specified for the key that the cursor passed instead of key input is performed. This is a way to ensure that words are selected. Figure 44 shows that the key input for 'boy' has been completed to input 'I am a boy.' ‘I an a not’ is displayed in the input window (44g). The reason why 'an' is displayed instead of 'am' is because 'an' is used more frequently than 'am', so 'an' is displayed in the input window among the predicted words, and in the case of 'boy', 'not' is better than 'boy'. Due to the high frequency of use, 'not' is displayed in the input window. To select the 'boy' you want to input, move the cursor (44f) from the central key area past the number 3 key (44c) to the word selection outer area (44f). When you return, input 'not' instead of key input. 'boy' is selected and entered in the window. Figure 45 shows a situation where 'boy' is selected and entered into the input window.

[3] Predictive sentence selection - joystick

Figure 46 shows the key input up to 'b', the first letter of the last word 'businessman', in order to input 'I am a businessman.' according to the sentence prediction input method of the present invention. 'I am a boy.', 'I am a businessman.', 'I am a banker.' Three predicted sentences are displayed. In the case of a joystick, the cursor cannot move outside the key area, so in order to press the 'sentence' key function key (46a) in Figure 46, which is a function key, the 'sentence' function must be assigned to a separate physical key, and this physical 'sentence' key must be When pressed, the 'sentence' function key 46a on the keyboard is clicked, and as shown in Figure 47, a predictive sentence is displayed on each letter key, allowing selection/input. When the prediction sentence is displayed on the key, move the joystick in the direction of the key (the key marked (II) in Figure 46) showing the prediction sentence you want to input, move the cursor to key (II), then return to the original center key and press 'I The input 'am a businessman.' is made. Figure 47 shows the result of 'I am a businessman.' being entered into the output window.

If, as seen in Example 45, two joysticks operate in conjunction with the keyboard and two cursors are used as shown in FIG. 38, one joystick is predicted in the same way as in item [1] of this embodiment. It is used to select words, and the other joystick is used to select predicted sentences in the same way as in item [1] of this embodiment, so that predicted sentences can be easily selected/entered without using separate function keys. Predictive characters are displayed in the input window (47b - a window that shows the current key input status) when selection is made, while predictive sentences are entered in the output window (47a) rather than the input window (47b) when selection/input is made. The final text input is performed.

[4] Predictive sentence selection - eye mouse

The method of selecting a predicted sentence using the eye mouse is the same as the method using the joystick. However, there is only a difference in the method of pressing the 'sentence' function key in Figure 46 to display the predicted sentence in the key area. That is, in the joystick method, the action of clicking the 'sentence' key (46a) on the keyboard is performed by pressing a separate hardware 'sentence' function key, whereas when using an eye mouse, the 'sentence' function shown in Figures 44 and 46 is performed. When you move the cursor with a key and return to the central key area, the predicted sentences in Figure 47 are displayed in the key area and can be selected/entered. In this state, if you move the cursor to the key (II) displaying the predicted sentence 'I am a businessman.' and then return to the central key area, 'I am a businessman.' will be entered in the output window.

As discussed up to Example 49, the text input method using the sentence prediction input method of the present invention is a method of easily and quickly entering text in cases where the user cannot see his or her hand or is inconvenient to operate the mouse cursor, such as with a remote control or eye mouse. is providing. With these features and advantages, the input method of the present invention enables smooth character input on VR, AR, and XR devices.

Example 49 In [1] and [2], we looked at a method of selecting words constituting a sentence from a list of predicted words during the key input process in the sentence prediction input method of the present invention. In this embodiment, a method of selecting/entering a word will be explained when all key inputs for sentence input have been made.

Figure 45 shows the state in which key input has been completed to input 'I am a boy.' In this case, the predicted word extracted from the database is 'an' in the case of 'am' and displayed in the input window, and 'not' in the case of 'boy' is extracted and displayed in the input window. In Example 49, the desired word may be selected from the predicted word list during key input, but as in the present embodiment, the desired word may be selected after all key input for sentence input has been completed. In this way, when all key inputs for sentence input are completed, pressing the 'voca' key 45a shown in Figure 45 (the pressing method is pressing the 'sentence' key 46a in Figure 46 described in Example 49) (Same as the method of giving), the mode is switched to sentence word editing mode, and as shown in FIG. 50, the constituent words of the sentence shown in the input window 49a are arranged on each key. Among the words shown in the input window, the second word 'an' must be changed to 'am', and the fourth word 'not' must be changed to 'boy'. How to change these words, the process of changing the second word 'an' to 'am' is as follows.

In Figure 49, when key (II) designated with 'an' is selected, the predicted word list corresponding to 'am', the second word of the sentence, is placed on keys (I) to (VI), as shown in Figure 50. Predicted word selection changes to mode. If you select the key (II) corresponding to 'am' in the predicted word selection mode where the predicted words are arranged like this, 'an' in the input window changes to 'am' and each key is a sentence in which the words that make up the sentence are arranged again. It returns to the word editing mode as shown in Figure 51. Before changing the sentence word in the prediction word selection mode, 'an' was assigned to the key (II) in Figure 49, which is the sentence word modification mode, but as a result of changing the prediction word in the prediction word selection mode, in Figure 51, the key (II) was You can see that 'am' is specified.

In Figure 50, which is the predicted word selection mode situation, there are only four predicted words corresponding to 'am', so the key (V) and key (VI) are empty. If there are more than 6 predicted words, you can select the key (VIII) marked with the word 'next' and place the remaining predicted words that are more than 6 in order on keys (I) to (VI). .

In the sentence word editing mode shown in Figure 51, the predicted word corresponding to 'boy' is still 'not', so this can be done in the same way as changing 'an' to 'am'. That is, after switching to the predicted word selection mode by selecting the key (IV) marked 'not' in Figure 51, and selecting the key marked 'boy' while the predicted words corresponding to 'boy' are displayed on each key, As shown in Figure 52, the completed sentence 'I am a boy' is obtained in the input window.

<Enter words not in the dictionary database>

In this embodiment, a method of inputting a word by entering the letters that make up the word will be explained when the word to be input is not in the prediction list. The word prediction input method (disambiguation input method), which constitutes the present invention, records dictionary words in a database and extracts predicted words according to key input. Words that are not actually registered in the dictionary are completed by entering one letter at a time like a regular keyboard. It has to be done. The method of entering a word that is not registered in the dictionary database will be explained using FIG. 50 as an example. Figure 50 shows a state in which the predicted word corresponding to 'am' is arranged in the key, as described in Example 50. If the word you want to input is not 'am' but 'fn', it is not among the predicted words, so to input 'fn', key (VII) shown in Figure 50 (marked as '<rev>', which is a predicted word) By selecting revise (meaning to revise/complement), the letters 'f' and 'n' corresponding to 'fn' are changed to 'spelling input mode'. The state changed to spelling input mode is shown in Figure 53.

When changing to spell input mode, letters are assigned to the keys instead of predicted words or sentences, allowing you to input one letter at a time like a regular keyboard. That is, in Figure 53, the letters 'a, s, d, f' assigned to the keys pressed during the key input process to input 'a', the first letter of 'am', are displayed on keys (VII), (II), (IV), It is specified in (VI), and a number is assigned to the key (V) for entering numbers. In addition to letters and numbers, you can also input symbols. If the word you want to input includes symbols, select key (III) to switch to symbol input mode and enter symbols like letters or numbers. Figure 53 shows the keyboard changed to symbol input mode. After completing the input of the symbol in the symbol input mode, select key (VII) (displayed as '<Back>', which means switching to the spelling input mode) to return to the spelling input mode shown in Figure 52 and complete the spelling input. I do it. Therefore, since the word to be input in Figure 52 is 'fn', if you select the key (VI) designated with 'f', the first letter of 'fn', the letter as shown in Figure 55 is entered so that you can input the next letter 'n'. 'b, n, m' are assigned to keys (II), (IV), and (V), and the number '5' is assigned to key (V). In this state, if the key (V) designated with 'n' is selected, character input is completed, and 'fn' can be seen displayed in the input window instead of 'am', as shown in Figure 56. And if you want to add a symbol to 'fn', you can add a symbol by selecting key (VII) (Designating 'End' means ending the spelling input mode.), and pressing the key marked 'End' (IV ) to exit the spelling input mode. And at the same time as terminating the spelling input mode, the entered 'fn' is registered in the dictionary database, and later 'fn' is also extracted as a predicted word and can be selected/entered from the predicted word list. A feature of symbol input in spell input mode is that symbols can be entered before or after letters, and symbols can even be entered continuously, completing the input of letters, numbers, and symbols equivalent to a general keyboard.

<How to add symbols to words that make up sentences>

Since the symbols are not visible on the keyboard for the sentence prediction input method shown in Figure 38, the symbols must be entered using a separate method. Proceeding with symbol input during the key input process reduces input efficiency due to changing the mode for character input and symbol input, so in this embodiment, symbol input is performed after all key input for sentence input is completed. It is a composition. In reality, symbols are placed before and after words as a function to assist the meaning of the word, so it is common to use them before the word is entered or after the word is keyed. Sometimes a symbol is embedded within a word. In the case of a word containing such a symbol, the process proceeds as in Example 51. If the symbol is not included in the word, the symbol can be entered according to the method of this embodiment.

In Example 51, when a symbol is included in a word, the symbol input process is explained in the spelling input process to input a word that is not registered in the dictionary database. In contrast, this embodiment is a process of entering symbols before and after the word after the key input for word input has been completed. As in the input process of 'I am a boy.' already shown in Example 49, when the 'voca' function key (49a) is selected when the key input for the word is completed, the words that make up the sentence as shown in Figure 49 (To be exact, recommended words that correspond to the words that make up the sentence) are displayed in keys (I) to (VI). In this state, if you select keys (I) to (VI), you enter the word modification stage. This is a modification mode for the words that make up the sentence, and the process of changing each word to another word in the predicted word list is an example. We looked at it in 50.

In this embodiment, in the situation shown in Figure 52 ('I am a boy'), select the key (VII) to which '<Sym>' is designated in and switch to symbol input mode to enter symbols before and after the words that make up the sentence. I would like to explain the process. When the key (VII) in Figure 52 is selected, the phrase '<Sym>' assigned to this key is activated, and if the key (VII) is selected once more in this state, it is deactivated and enters symbol input mode. It is desirable to display the change in status by changing the background color or modifying the text such as '<Sym+>' so that the user can recognize the activation status. As shown in Figure 57, when the symbol function key is activated and the keys (I) to (VI) designated with the words that make up the sentence are selected, the symbol input mode can be entered as shown in Figure 58 and the symbol can be input. In Figure 58, the key (I) in Figure 57, which is in the symbol input mode, can be selected to input symbols before and after the first word of the sentence, 'I'. If the key (I) is selected in this state, the symbol ['] is entered in front of the word 'I'. ' It shows the input state at the front. If the cursor passes the key input outer boundary 42b and returns to the center key, it exits the symbol input mode and returns to the sentence word editing mode. To enter [.] and ['] after 'boy', select the key (VII) to which '<Sym>' is assigned to switch to symbol input mode, and then enter the phrase 'to switch the left input mode of the symbol to the right input mode. When the key (VII) with <--' specified is selected, the text is changed to '-->' and Figure 60 shows the state converted to the right input mode. To enter the period symbol [.], select key (VI) and then select key (I) to complete 'I am a boy.' When the symbol input is completed in this way and the cursor moves beyond the outer boundary of the key input and returns to the central key area, it switches to the sentence word correction mode and completes the input, as shown in FIG. 62. The above has explained the case where sentence prediction input is performed in connection with the eye mouse. If sentence prediction input is performed in conjunction with a joystick, it is the same as an eye mouse, and only in the process of switching from symbol input mode to sentence word editing mode, the cursor does not need to leave the outer boundary of the key input, with the cursor located in the center key area. Just press the joystick. In other words, the joystick cannot perform mode conversion by moving the cursor because the movement of the cursor is limited only within the outer boundary of the key input, but instead, it has a push switch function, so mode conversion can be performed using the pressed state. . Therefore, by linking both the joystick and the eye mouse, the sentence prediction input method of the present invention can be performed with a keyboard having nine keys, as shown in Figure 42. In the case of a joystick, if there is no push function, just like the sentence function key in Example 49 [3], a physical button to handle the push function can be designated to take charge of that function.

<Conversion of uppercase/lowercase letters of English words>

The sentence prediction input method of the present invention does not have an uppercase/lowercase conversion key found on an English keyboard. Therefore, this embodiment provides a method for converting uppercase and lowercase letters. In the present invention, uppercase and lowercase letters are not distinguished when keying, so the uppercase/lowercase conversion of the word to be input is performed after keying. If this method is applied to the sentence prediction input method in conjunction with the eye mouse, it is as follows.

As shown in Figure 57, in a state where the word to be converted to uppercase ('am') is not switched from the sentence word editing mode to the symbol input mode, the cursor passes the designated key, passes the key input outer border, and returns to the center key area. When it comes, the first letter of the word is capitalized ('Am' - Figure 63). (Performs the function of the 'Shift' key on a regular keyboard). With the first letter capitalized, the cursor passes the key where the word is designated and goes outside the key input. When you pass the border and return to the central key area, all letters of the word assigned to this key will be capitalized ('AM' - Figure 64) (performing the 'Caps' key function on a regular keyboard), and all words will be capitalized when the cursor moves. If you pass the designated key, pass the outer border of the key input, and return to the central key area, all letters are lowercase again ('am' - Figure 64).

When the above uppercase/lowercase switching method is linked to a joystick and the sentence prediction input method is performed, in the case of an eye mouse, the cursor passes the outer border of the key input and returns to the center key area, which is replaced by the joystick's press function. If the joystick does not have a pressing function, a separate physical button can be used to perform the pressing function and the uppercase/lowercase switching function can be used.

As discussed up to Example 53, the advantage of the sentence prediction input method of the present invention in promoting convenience of input as a small device will greatly contribute to enabling users to communicate without inconvenience as an input system in the coming VR era.

Claims (9)

In the word prediction input method, where a number of letters are assigned to one key, and these keys are pressed to extract and input words made from a combination of the letters assigned to these keys according to the key input sequence, not only letters but also symbols are used as letters. An input system that implements a predictive input method that assigns multiple symbols to one key and extracts not only words but also phrases and sentences containing the symbols specified in the key by pressing the key only once without distinguishing between these symbols. claim In 1, an input system that implements a phrase prediction input method that includes blank characters (space) among the symbols input by the phrase prediction input method. In claim 1, if there is no sentence corresponding to the ambiguous key input for sentence input, in the process of registering the sentence according to the entered ambiguous key sequence, word prediction input is performed for the remainder except for the ambiguous key input corresponding to symbols and space characters. An input system that implements a method of registering sentences by applying
In claim 1, an input system that implements a method of extracting the sentence as a predicted sentence from the database and allowing selection/input even if only a part of each word constituting the sentence to be entered is entered when entering an ambiguous key for sentence input.
In claim 1, a predictive input system for an input device with limited keys by distinguishing between letters and symbols by assigning a symbol to one key.
In claim 2, the meaning of 'space' is a predictive input system that distinguishes between 'sentence-space', which distinguishes words in a sentence, and 'search-space', which distinguishes search words or clauses during a search.
In claim 1, a predictive input system that has a key arrangement that allows input at once without using expansion keys by switching between the 'space' key and 'symbol' key arranged on the keyboard depending on the input environment.
In claim 1, a sentence prediction input system that utilizes a word prediction input method that displays predicted words on the keyboard and allows selection during the process of registering in the database because there are no sentences or clauses to be searched after key input for sentence prediction input is made.
claim In 1, in the case of Korean predictive input, a grammatical blank key is specified, and in the case of verbs and verbs, the verb and stem are extracted into a dictionary database separately from the particle and the ending, and are combined with the separately extracted particle and ending prediction list.