KR101355284B1 - Method for Recommending Words and Completing Sentences in Touch Screen Devices - Google Patents

Abstract

According to the present invention, a character is input using a virtual keyboard, and a selected recommendation word is displayed based on the input character, and a touch screen unit for selecting the displayed recommendation word is to be compared based on the input character. A word extraction unit for extracting en-gram data and selecting a recommendation word in order of high frequency of appearance and outputting the recommended word to the touch screen unit; and providing the en-gram data in connection with the recommendation word extracting unit, The present invention provides a word recommendation and sentence completion system and a method in a touch screen environment including an en-gram DB construction unit for extracting an en-gram to build an en-gram database.
According to the word recommendation and sentence completion system and method in the touch screen environment of the present invention, based on the en-gram data constructed by selecting only a specific domain in a highly reliable web document, the sequence of words entered by the user and the corresponding syllables are matched. Input the correct word with the minimum key input by selecting the en-gram data to be selected, outputting the recommended words in order of frequently used words, that is, frequently used words, and selecting the recommended words that match the user's intention. And complete the sentence easily.

Description

Method for Recommending Words and Completing Sentences in Touch Screen Devices}

The present invention relates to a word recommendation and sentence completion system and method in a touch screen environment, and more particularly, to a system and method for recommending words and completing sentences when a character is input in a device to which a touch screen is applied. It is about.

Recently, due to the rapid development of smart phones and the spread of devices with various touch screens, the way of inputting words by physically pressing the keyboard has been changed to the environment for inputting words on the touch screen keyboard with the virtual keyboard.

Such a virtual keyboard uses a qwerty keyboard to input words, but typos frequently occur due to the size of a small terminal, lack of physical boundaries, sensor problems on the touch screen, and the like. In addition, due to the difficulty of pressing a virtual button of a small screen, the character input time is lengthened, making the user feel uncomfortable. In addition, people suffering from physical discomfort, among other things, the need to enter the desired word at least by keystrokes.

Korean Publication Patent: 10-2008-0039009 (Published 2008. 05.07)

Korea Patent Registration: 10-0892003 (Notice date 2009. 04.07)

SUMMARY OF THE INVENTION The present invention has been made to solve the conventional problems,

An object of the present invention is to select a sequence of words entered by the user and the en-gram data corresponding to the corresponding syllables based on the en-gram data constructed by selecting only a specific domain in a highly reliable web document, and a word having a high frequency of occurrence. Suggested words are output in order, and by selecting the recommended words that match the user's intention, the word recommendation and sentence completion system and the sentence completion system in a touch screen environment that can input the correct word with a minimum key input and can easily complete the sentence. To provide a way.

In the touch screen environment of the present invention according to an aspect provided to achieve the above object, the word recommendation and sentence completion system extracts en-gram data to be compared based on words and syllables input to the touch screen. In the extracted en-gram data, a word is recommended and output in the order of the high frequency of appearance, and the sentence is completed by selecting the output recommendation word.

The system includes a touch screen unit for inputting letters and selecting recommended words; A recommendation word extracting unit extracting en-gram data to be compared based on the input characters and recommending and outputting words in order of appearance frequency; And an en-gram DB construction unit for constructing an en-gram database for providing en-gram data to the recommendation word extracting unit.

The touch screen unit may include a text input unit for inputting a character, an input word output unit for outputting an input character and a completed sentence, and a recommended word output unit for outputting a recommended word.

The en-gram DB construction unit is divided into unigram data consisting of one word, bigram data consisting of a combination of two consecutive words, and trigram data consisting of a combination of three consecutive words, and calculating and storing the frequency of occurrence of each word. It is characterized by one.

The recommended word extracting unit determines whether the first syllable of the input character belongs to the first, second, or third words, and if the first syllable is the first syllable, searches the unigram data and searches the first syllable of the first word. If the syllable is searched in the bigram data, and the first syllable of the third word is searched in the trigram data.

In addition, the en-gram DB construction unit selects only a specific domain from the web document, extracts the abstract text of the selected specific domain, characterized in that the construction of the en-gram data through a preprocessing process to remove the special characters.

According to another aspect of the present invention, in the touch screen environment of the present invention, the word recommendation and sentence completion system inputs a character using a virtual keyboard, selects a selected recommendation word based on the input character, and selects a displayed recommendation word. Touch screen unit that can be; A recommendation word extraction unit configured to extract en-gram data to be compared based on the input characters, select recommended words in order of appearance frequency, and output the selected words to the touch screen unit; And an en-gram DB constructing unit connected to the recommendation word extracting unit to provide en-gram data and extracting the en-gram from the purified specific domain to build an en-gram database.

The touch screen unit may include a text input unit for inputting a character using a virtual keyboard, a recommendation word output unit for outputting and displaying a recommendation word, and selecting a displayed recommendation word, and an input for outputting and displaying the character input from the text input unit. Characterized in that the word output unit.

Here, when a word is selected based on the recommended word displayed on the recommended word output unit, the selected word is displayed on the input word output unit.

The recommendation word extracting unit selects an en-gram to be compared based on the input word and syllable, and recommends selecting the recommendation word in order of high frequency of appearance among the selected en-gram data. Characterized in that the word selection unit.

The gramgram DB construction unit includes a single word, a unigram data unit configured to calculate and store the frequency of each word, a combination of two consecutive words, and a gramgram data unit configured to calculate and store the frequency of each word; Combination of three consecutive words, characterized in that consisting of a trigram data unit for calculating and storing the frequency of each word.

In addition, the engram DB construction unit includes a preprocessor for simplifying a large amount of web documents, the preprocessor is a specific domain selection unit for selecting a major area of interest, an extraction unit for extracting the abstract text, filtering to remove special characters It is characterized by consisting of wealth.

According to the word recommendation and sentence completion system and method in the touch screen environment of the present invention, based on the en-gram data constructed by selecting only a specific domain in a highly reliable web document, the sequence of words entered by the user and the corresponding syllables are matched. Input the correct word with the minimum key input by selecting the en-gram data to be selected, outputting the recommended words in order of frequently used words, that is, frequently used words, and selecting the recommended words that match the user's intention. And the effect is easy to complete the sentence.

In addition, by entering the correct word with a minimum key input and easy to complete the sentence, it is possible to prevent typos from occurring, shorten the input time of characters, and improve user convenience.

1 is a block diagram illustrating a word recommendation and sentence completion system in a touch screen environment according to the present invention.
2 is a conceptual diagram illustrating an en-gram construction and a word recommendation system using the same according to a specific domain selection in a word recommendation and sentence completion system in a touch screen environment according to the present invention.
3 is a flowchart illustrating a word recommendation and sentence completion method in a touch screen environment according to the present invention.
4 is a flowchart illustrating a process of extracting a recommended word in a word recommendation and sentence completion method in a touch screen environment according to the present invention.

These and other objects, features and other advantages of the present invention will become more apparent by describing in detail preferred embodiments of the present invention with reference to the accompanying drawings. Hereinafter, a word recommendation and sentence completion system and a method thereof in a touch screen environment will be described in detail with reference to the accompanying drawings. For purposes of this specification, like reference numerals in the drawings denote like elements unless otherwise indicated.

1 is a block diagram illustrating a word recommendation and sentence completion system in a touch screen environment according to the present invention, Figure 2 is an en-gram according to the selection of a specific domain in the word recommendation and sentence completion system in a touch screen environment according to the present invention A conceptual diagram showing the construction and the word recommendation system using the same.

In the touch screen environment of the present invention, the word recommendation and sentence completion system extracts the en-gram data to be compared based on the words and syllables input on the touch screen, and in the order of appearance frequency among the extracted en-gram data. Suggest a word and print it, and complete the sentence by selecting the suggested word.

As shown in FIG. 1 and FIG. 2, the system of the present invention includes a touch screen unit 100, a word extraction unit 200, and an en-gram DB construction unit 300.

The touch screen unit 100 refers to a screen window in which an input is made by a touch in a device to which a touch screen is applied. The touch screen unit 100 is a text input unit 110, the virtual keyboard is applied to the character is input, and the recommended word output unit 120 and the text input unit for outputting and displaying the recommended words and select the displayed recommended words; The character input at 110 is outputted and displayed, and when the recommended word is selected at the recommended word output unit 120, the selected recommended word is output and displayed at the input word output unit 130. As for the virtual keyboard of 110, a qwerty keyboard is preferably used.

The recommended word extracting unit 200 includes a comparison en-gram determining unit 210 and a recommended word selecting unit 220.

The comparison en-gram determination unit 210 extracts and selects the en-gram data to be compared from the en-gram DB construction unit 300 based on the words input from the text input unit 110 and the syllables.

In addition, the recommendation word selecting unit 220 selects the recommendation words in order of high frequency of appearance, that is, frequently used words, among the engram data selected by the comparison engram determining unit 210, and selects the selected recommendation. The word is output to the recommended word output unit 120 of the touch screen unit 100.

The en-gram DB building unit 300 simplifies a large amount of web documents 310 through the preprocessor 320. The preprocessor 320 includes a specific domain selector 321 for selecting a main field of interest, an abstract sentence extractor 322 for extracting an abstract sentence, and a filtering unit 323 for removing special characters.

The specific domain selector 321 first selects a specific domain having excellent reliability. That is, select a specific domain that selects a web document in a sentence that is structurally clear, such as Wikipedia or news, and a web document in a major area of interest. The web document 310 is described in various fields, and there is a need to focus on a specific field. The reason for this is that since the types of words used are different according to the field, the web documents necessary for constructing an engram are selected as specific domains.

The abstract sentence extractor 322 extracts the abstract sentence from the selected specific domain. Especially, in the case of Wikipedia documents, the abstract text is the most important part.

The filtering unit 323 removes all special characters such as quotation marks and commas from the extracted abstract text. In the web document 310, unnecessary words and special characters that do not have meaning frequently appear. Unnecessary words and special characters can play a role to make an error in the process of building the correct en-gram data, so the special character is removed to build the en-gram database unit 330.

The engram database unit 330 constructed through the preprocessing unit includes a unigram data unit 331 consisting of one word and calculating and storing a frequency of each word, and a combination of two consecutive words, A bigram data unit 332 that calculates and stores a frequency and a trigram data unit 333 that combines three consecutive words and calculates and stores a frequency of each word.

In the touch screen environment having the above configuration, the word recommendation and sentence completion system is a comparison diagram of the recommendation word extracting unit 200 when the first syllable of a word is input from the text input unit 110 of the touch screen unit 100. The determination unit 210 determines whether the first syllable input belongs to the first, second, or third words, and if the input syllable is the first syllable 211 of the first word, the engram DB construction unit ( The unigram starting from the input syllable is searched in the unigram data 331 of 300. For example, when an English syllable T is input, the word beginning with T is searched for and extracted from unigram data. The recommendation word selecting unit 220 sorts in order of the frequency of appearance of the words starting with T and outputs the sorted word to the recommendation word output unit 120. In this case, if the recommendation word does not exist, the next syllable is input to select a more accurate recommendation word. For example, when a syllable h following an English syllable T is input, the comparison en-gram determination unit 210 searches for a unigram starting with Th, and the recommendation word selecting unit 220 shows an occurrence frequency among words starting with Th. The numbers are sorted in ascending order and output to the recommended word output unit 120. Subsequently, when a word matching the user's intention is selected based on the recommendation word output to the recommendation word output unit 129 of the touch screen unit, the selected word is output and displayed on the input word output unit 130.

In addition, when the syllable input through the text input unit 110 is the first syllable 212 of the second word, the comparison en-gram determination unit 210 of the recommendation word extracting unit is a bigram of the en-gram DB construction unit 300. In the bigram data 332, the first word inputted and the first user's first input syllable are searched. For example, when the first word previously input is This and the first syllable currently input is i, the first word of the data is searched for this first and the second word starts with i. The recommendation word selecting unit 220 outputs the recommended word selecting unit 220 to the recommendation word output unit 120 by sorting in order of the frequency of appearance of the viagrams starting with This i. That is, this information, This interest, This is, This investigation, etc. may appear as an example of the current process. At this time, if there is no recommendation word starting with This i, the correct syllable word is selected by receiving the syllable after This i. Subsequently, when the user selects the word information corresponding to the intention among the information, This interest, This is, and This investigation, which is output to the recommendation word output unit 120, the selected This information is input to the input word output unit 130. Is displayed.

In addition, when the syllable input through the text input unit 110 is the first syllable 213 of the third word, the comparison en-gram determination unit 210 of the recommendation word extracting unit is a tri-section of the en-gram DB construction unit 300. The trigram data 333 searches for a trigram starting with the first word, the second word, and the first syllable currently input. For example, when the second word previously input is This information and the first syllable currently input is w, the comparison engram determining unit 210 determines that two words of the trigram data are This information and the third word. Searches for trigrams starting with w. In addition, the recommendation word selecting unit 220 sorts the order of appearance frequency among the trigrams starting with This information w and outputs the result to the recommended word output unit 120. At this time, if there is no recommendation word starting with This information w, a more accurate recommendation word is selected by receiving the syllable following this information w. Subsequently, when the user selects a word that matches the intention based on the recommendation word output to the recommendation word output unit 120, the selected word is output and displayed on the input word output unit 130.

In the system of the present invention as described above, a word is recommended based on a word and a syllable input when a character is input through the text input unit 110 in a touch screen environment. As the sentence is completed, you can enter the correct word and complete the sentence easily with a minimum of keystrokes.

 3 is a flowchart illustrating a word recommendation and sentence completion method in a touch screen environment according to the present invention, and FIG. 4 is a flowchart illustrating a process of extracting a recommended word in the word recommendation and sentence completion method in a touch screen environment according to the present invention. to be.

3 and 4, the word recommendation and sentence completion method in the touch screen environment of the present invention comprises the steps of inputting a character using a virtual keyboard in a touch screen environment (S100); Extracting en-gram data to be compared based on the input character (S200); Selecting and outputting a recommended word in order of appearance frequency among the extracted en-gram data (S300); And completing a sentence by selecting a word that matches the intention of the user based on the output of the recommended word (S400).

In step S100, words and syllables are input using a virtual keyboard of a touch screen environment.

In the step S200, the en-gram data is selected in a specific domain of the web document of the main field of interest that is structurally clear and the web document and the en-gram of a large amount of web documents, extracting the abstract sentence, and extracting special characters. The preprocessing process removes all the data and extracts them from the en-gram database that constructs the en-gram data.

And, the en-gram database is constructed by dividing into unigram data consisting of one word, bigram data, which is a combination of two consecutive words, and trigram data, which is a combination of three consecutive words. The number of occurrences of each word is calculated and stored.

As shown in FIG. 4, the extraction of en-gram data in step S200 is performed based on the word and syllable input in step S100, and the first syllable input first to select the type of en-gram data to be compared is the first word. In operation S211, the second word S221 and the third word S231 are determined.

As a result of the determination, when the input syllable is the first syllable of the first word (S211), the unigram starting with the input syllable is searched from the unigram data of the en-gram database (S212). In this case, if the syllable input does not exist, the next syllable is input to select a more accurate recommendation word. For example, when the first syllable of the first word is T, the unigram data is searched for a word starting with T, and it is determined whether a word starting with T exists (S241). When the word starting with the search result T does not exist (S241), the next syllable is input (S242). For example, a word beginning with Th is searched for in a unigram data by receiving a syllable h after the first syllable T.

If the input syllable is the first syllable of the second word (S221), the first word input from the bigram data of the engram database and the firstgram currently input with the first syllable are searched. (S222). For example, when the first word previously input is This and the first syllable currently input is i, the first word of the data is searched for this first and the second word starts with i. That is, this information, This interest, This is, This investigation, etc. may appear as examples of the current process. At this time, if there is no recommendation word starting with This i, the syllable is input after the This i and re-searched (S242).

In addition, when the input syllable is the first syllable of the third word (S231), the first word, the second word, and the first syllable currently input from the trigram data of the engram database are determined. Search for the starting trigram (S232). For example, when the second word entered is This information and the first syllable currently entered is w, the two words in the trigram data are This information and the third word starts with w. . At this time, if there is no recommendation word starting with This information w (S241), the next syllable is input and then re-searched (S242).

In step S300, when the search result recommendation word in step S200 exists, the search word is selected and output in order of appearance frequency.

In operation S400, the word corresponding to the user's intention is selected based on the output of the recommended word. The selected words are outputted and displayed on the touch screen as completed words and sentences.

As described above, in the touch screen environment of the present invention, the word recommendation and sentence completion method selects a specific domain from a highly reliable web document based on en-gram data that is input by the user, and matches the sequence number of the word entered by the user and the corresponding syllable. By selecting the gram data and outputting the word on the touch screen mainly on words with high frequency of occurrence, that is, frequently used words, it is possible to input the correct word with minimum key input and easily complete the sentence by selecting the word that matches the user's intention. It becomes possible.

Although the preferred embodiments of the present invention have been described, the present invention is not limited to the specific embodiments described above. It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit or scope of the appended claims, And equivalents may be resorted to as falling within the scope of the invention.

100: touch screen unit 110: text input unit
120: recommended word output unit 130: input word output unit
200: recommended word extracting unit 210: comparison en-gram determination unit
220: recommended word selection unit 300: N-gram building unit
310: web document 320: preprocessor
321: specific domain selector 322: abstract sentence extractor
323: Filtering unit 331: Unigram data unit
332: Baigram data portion 333: Trigram data portion

Claims (20)

delete delete delete delete delete delete delete delete delete delete delete delete delete Inputting text using a virtual keyboard in a touch screen environment;
Extracting en-gram data to be compared based on the input characters, wherein the en-gram data is extracted from an en-gram database constructed by extracting en-grams from a specific purified domain;
Selecting and outputting a recommended word in order of appearance frequency among the extracted en-gram data; And
Comprising the step of completing the sentence by selecting the word based on the output of the recommended word,
Extracting en-gram data to be compared based on the input characters, wherein the en-gram data is extracted from an en-gram database constructed by extracting the en-gram from a specific purified domain;
The en-gram database constructs en-gram data through a preprocessing process that simplifies a large amount of web documents, but the pre-processing process selects web documents of major areas of interest necessary for en-gram construction as a specific domain. To extract all of the special characters.
The en-gram database is constructed by dividing unigram data composed of one word, bigram data composed of two consecutive words, and trigram data composed of three consecutive words, and calculating the frequency of occurrence of each word. Save it,
The en-gram data extraction determines whether the first syllable of the input word is the first syllable of the first word, retrieves the unigram data when the first syllable is the first syllable, and the bigram data when the first syllable is the first syllable of the second word. , The first syllable of the third word, extracted by searching the trigram data,
If the recommended word is not determined by the first syllable of the input character, the word recommendation and sentence completion method in a touch screen environment characterized in that the second syllable is input and re-search.
delete delete delete delete delete delete

Images