KR20040046166A - Korean input device and method thereof of cellular phone - Google Patents

Abstract

PURPOSE: A non-deterministic Hangul inputting apparatus and method using a syllable context probability model and a table-based parsing are provided to considerably reduce the number of times of manipulating buttons to input characters in a mobile terminal. CONSTITUTION: A contract type Hangul key board unit(10) generates a code assigned to a button when the button is manipulated. A button stream parsing unit(20) processes inputted buttons and generates available candidates and order of the candidates. A button stream/syllable corresponding table(30) is data used for effectively searching syllables. Syllable context probability data(40) are used to calculate the order of candidates according to probability. A parsing table(50) is used to store candidate syllable streams and erase re-calculation when a button is additionally inputted. A candidate selecting/processing unit(60) determines a syllable stream or changes a page of a candidate list. A screen output unit(70) outputs suitable content on an LCD with reference to the parsing table(50).

Description

Nondeterministic Hangul input device and method using syllable context probability model and table-based parsing

The present invention relates to a non-deterministic Hangul input device and method using a syllable context probability model and table-based parsing.

In particular, the present invention relates to a technology that can significantly reduce the number of button manipulation by applying a method for solving the problem of the conventional non-deterministic Hangul input method in a Hangul input environment using a reduced Hangul keyboard.

Recently, with the expansion of mobile communication terminals and the provision of various functions and services, the functions of mobile communication terminals have moved beyond the scope of simple voice communication devices. Among the additional functions and services of the mobile communication terminal, the transmission of short messages, electronic notebook management, schedule management, and chat, which are frequently used, are based on input and output of text. In a small electronic device such as a mobile communication terminal requiring text input, a reduced Hangul keyboard is used due to space constraints.

As the popularity of mobile communication terminals increases and the need for Hangul input increases, various Hangul input methods using a reduced Hangul keyboard have been devised. Among them, Korean Patent Publication No. 1997-0002558, Korean Patent Publication No. 1997-0072848, Korean Patent Publication No. 1999-0067835, and Korean Patent Publication No. 2000-0049346, Republic of Korea Patent Publication No. 2000-0050253, Republic of Korea Patent Publication No. 2002-0001918, Republic of Korea Patent Publication No. 2002-0053689.

The Republic of Korea Patent Publication No. 1997-0002558, Republic of Korea Patent Publication No. 2002-0001918, Republic of Korea Patent Publication No. 2000-0050253 and Republic of Korea Patent Publication No. 1999-0067835 have adopted a multiple typing method, each Except for Korean Patent Publication No. 2000-0050253, which presses a function button at the boundary of a syllable, the same button ambiguity is generated, which increases the input time.

For example, the Korean Patent Publication No. 1997-0002558 for the same button ambiguity is as follows. Figure 2 schematically shows the keyboard layout for the technical content described in the Republic of Korea Patent Publication No. 1997-0002558. In this technical description, the syllable sequence "Country" and "Kukka" both enter buttons '4', '3', '2', '4', '4', '1', and '2' sequentially. It is defined to press. However, if you press the input buttons '4', '3', '2', '4', '4', '1', and '2' in order to actually enter "Country", it will be incorrectly recognized as "Kuk". Therefore, to solve this problem, after pressing '4', '3', '2', or '4', an additional syllable confirmation button is additionally pressed or waited for a specified time or more.

Meanwhile, conventional Hangul input methods using the reduced Hangul keyboard may be classified into a deterministic phoneme input method and a nondeterministic phoneme input method according to the ambiguity processing method. The deterministic phoneme input method has the advantage of fast processing for button input and less required storage space because no processing for ambiguity is required, but there are a large number of dependent phonemes and two or more button operations are required for dependent phonemes. Therefore, the input speed is slow. On the other hand, the nondeterministic phoneme input method having a ambiguity resolution rate of 100% has an advantage of inputting Korean text using only a few button operations compared to the deterministic phoneme input method. However, in reality, since the ambiguity resolution rate is less than 100% in the non-deterministic input method, an additional button operation is required to correct an error occurring during the ambiguity resolution process, and when the ambiguity resolution rate is low, The number of button manipulations may require more button manipulations than the deterministic phoneme input method.

The Republic of Korea Patent Publication No. 1997-0002558, Republic of Korea Patent Publication No. 2000-0049346, Republic of Korea Patent Publication No. 2002-0001918 and Republic of Korea Patent Publication No. 2002-0053689 are widely used deterministic phoneme input methods, The Republic of Korea Patent Publication No. 1997-0072848, Republic of Korea Patent Publication No. 1999-0067835, Republic of Korea Patent Publication No. 2000-0050253 are non-deterministic phoneme input methods that have been put to practical use. Table 1 shows the results of comparing these methods in terms of the number of button manipulations. Table 1 compares the number of button operations required to enter a document set consisting of 21,062,496 Korean words containing a total of 62,098,517 Hangul syllables compared to the number of button operations on a QWERTY 2 keyboard Hangul keyboard commonly used on personal computers. It is shown.

Patent Publication Number Button operation count Average number of button operations per syllable Growth rate of quality keyboard 1997-0002558 265,268,567 4.272 46.35% 2000-0049346 253,458,691 4.082 39.84% 2002-0053689 232,937,763 3.751 28.50% 1997-0072848 (a) 211,335,984 3.403 16.58% 1997-0072848 (b) 234,261,741 3.773 29.26% 1999-0067835 224,258,283 3.611 23.71% 2000-0050253 256,520,316 4.131 41.52% Quality two bee type keyboard 181,254,776 2.919 0.00%

In Table 1, the Republic of Korea Patent Publication No. 2002-0053689 was performed under the assumption that it does not support simultaneous pressing of buttons. The reason for this is that most mobile terminals that are currently commercialized do not recognize that two or more buttons are pressed at the same time. Another reason is that the input speed may be faster than when the buttons are simultaneously pressed. This is because the number of button operations is almost the same.

In Table 1, the method of Korean Patent Publication No. 1997-0072848 has two results, 1997-0072848 (a) and 1997-0072848 (b), as follows. 3 is a view schematically showing the keyboard layout for the technical content described in the Republic of Korea Patent Publication No. 1997-0072848. In the above method, the constraints according to the Hangul discipleship principle, the constraint that only the first syllable or complete syllable can be input, the constraint that the syllable boundary ambiguity or the phoneme classification ambiguity are generated, and the syllable division button after the syllable boundary ambiguity are generated When you enter a vowel, you eliminate the phonetic ambiguity by restricting you to press a function button that indicates it. According to the above method, it is not necessary to press the syllable separator button to input a syllable string including the non-spoken Hangul syllable in a situation where ambiguity does not occur. For example, if you enter a string called "addition or decrement", because the syllable boundary ambiguity occurs, the button string "1" for syllable "A", "3" and the button string "1" for syllable "sen", You should press the syllable separator button '*' between '3' and '5'. However, when inputting "monster", it is not necessary to press the syllable separator button "*" because ambiguity is resolved by the above constraints and only one candidate string exists. However, since it is difficult for the user to determine whether the current situation is a ambiguity situation or not, in reality, most symptom-free Hangul syllables are input, and then the syllable division button is pressed. In Table 1, 1997-0072848 (a) is an experiment result in assuming that the user knows all the ambiguities and presses the syllable division button only when it is absolutely necessary. This is the result of assuming that the syllable division button is pressed after the final Hangul syllable.

In Table 1, 1999-0067835 shows only experimental results for a syllable unit ambiguity resolution method among methods created in Korean Patent Publication No. 1999-0067835. The reason is that the Korean Patent Publication No. 1999-0067835 proposes a syllable unit ambiguity resolution method and a word unit ambiguity resolution method. This is because it is impossible to apply the word unit ambiguity resolution method as it is for various Korean characters.

As shown in Table 1, 1997-0002558, 2000-0049346, and 2002-0053689, which use the deterministic phoneme input method, require about one more button operation per syllable compared to the quality keyboard, which is the number of button operations. Considering this alone, it means that the input time is 38% longer on average. Since this is a structural problem of the deterministic phoneme input method using the reduced Hangul keyboard, even if the keyboard layout method is changed, there is a limit in reducing the number of button manipulations.

In Table 1, although 2000-0050253 is a nondeterministic phoneme input method, it can be seen that there is little advantage in terms of the number of button manipulations compared to the deterministic phoneme input method. The reason is that the method adopts syllable unit ambiguity resolution method and it is defined to press syllable separator button '#' at the end of every syllable, so the number of button manipulations for phoneme input is small, but the number of button manipulations for ambiguity resolution is relatively small. Because there are many. In addition, the number of buttons used to resolve ambiguities is low, but the ambiguity resolution rate is low, and the number of times the candidate selection button '*' is pressed during the ambiguity resolution process also increases the number of button manipulations.

In Table 1, although 1997-0072848 shows a relatively better performance than the deterministic phoneme input method, it has some problems as follows. First, when a user inputs an agnostic Hangul syllable and cannot determine whether ambiguity occurs, the user must press a syllable classification button. Second, if the user cannot determine that ambiguity does not occur when entering the vowel vowel, the button '#' indicating the continuous vowel input should be pressed. Third, if the last syllable can be combined with the previous syllable's finality and the previous syllable's finality can be a double consonant, the syllable distinction button should be pressed. For example, in the above method, to input the syllable string "foot", enter the button strings "6", "3", and "4" corresponding to "foot", then press the syllable separator button "*" and then " And press "1", "5", and "3". If you press '6', '3', '4', '1', '5', or '3', syllable boundary ambiguity occurs and it is recognized as "bright". As described above, the above method may cause confusion of the user depending on whether ambiguity occurs or not, so that the number of screen references (confirmation) increases, resulting in a smaller number of button manipulations compared to a deterministic input method, but the time required for input is reduced. There are disadvantages that can be more.

In Table 1, 1999-0067835 has the advantage of relatively low number of button operations. However, since the ambiguity resolution rate is lowered and the ambiguity is resolved in units of syllables, the user has to check the syllable input result through the screen reference for each syllable input.

As shown in Table 1, the nondeterministic phoneme input method has a feature of reducing the number of button manipulations compared to the deterministic phoneme input method. However, if the ambiguity resolution rate is low, the number of button manipulations may be increased. In the case of using the syllable unit ambiguity resolution method, the input time is increased because the user frequently needs to refer to the screen. Therefore, in order to reduce the number of button operations and the number of screen references at the same time, it is necessary to increase the ambiguity resolution unit and to increase the ambiguity resolution rate.

Increasing the ambiguity resolution unit and increasing the ambiguity resolution rate can significantly reduce the number of button manipulations compared to the deterministic phoneme input method. However, some practical problems must be solved in order to solve these two tasks.

The reason for using the syllable unit ambiguity resolution method in the conventional deterministic phoneme input method is that as the ambiguity resolution unit increases, the amount of data required for ambiguity resolution increases, making it difficult to transplant to an electronic device using a reduced Hangul keyboard. Because. For example, in the syllable unit ambiguity resolution method using probability, only the probability data of 2,350 Korean syllables included in the KSC 5601 character set need to be maintained. However, if the ambiguity resolution unit is extended to two syllables in the same way, probability data for 5,522,500 syllable combinations corresponding to 2,350 squares will be needed. It becomes impossible. Of course, instead of using the probabilities for 5,522,500 syllable combinations, you can use a subset of these probabilities or use more simplified data, but in that case, the ambiguity resolution will be lowered and the number of button manipulations will increase. Therefore, there is a need for a method that can increase the ambiguity resolution unit without reducing the ambiguity resolution rate using a small amount of data.

In general, the more ambiguity resolution techniques reflect linguistic phenomena, the higher the ambiguity resolution rate. However, the more complicated the ambiguity resolution model is, the more complicated the computational complexity is to reflect the linguistic phenomenon. Therefore, when the user presses a button, the response time of reflecting the result on the LCD may be slow.

In summary, if a Korean input method using a non-deterministic phoneme input method that has low computational complexity and a small amount of data but exhibits a high ambiguity resolution rate is devised, the number of button manipulations can be drastically reduced compared to conventional techniques. It can be seen that there is.

The present invention solves the ambiguity of the Hangul input method of the nondeterministic phoneme input method with the keyboard layout and syllable context probability model considering the contiguous distribution of the phonemes, and at the same time reduce the complexity of calculation and save the data storage space. The purpose is to provide a non-deterministic Hangul input device and method using a syllable context probability model and table-based parsing.

Another object of the present invention is to use a button sequence / syllable correspondence table in order to reduce internal code conversion time, and to reduce computational complexity by using table-based parsing to eliminate probabilities.

Still another object of the present invention is to enable a user to input Korean characters on a device having a reduced size keyboard more quickly and conveniently.

The present invention for achieving the above object is a phoneme button in which one or more Hangul phonemes are arranged in a Jamo mixture, a function button that is pressed to input dependent phonemes, and a candidate syllable string currently output to the liquid crystal display is a syllable string to be input. A reduced Hangul keyboard portion including a confirming button for confirming, a change button for outputting a next-order candidate to the liquid crystal display, a candidate selecting button for moving to the candidate selection mode, and a delete button for canceling the most recently inputted phoneme button; and Provided is a non-deterministic phoneme input method of a Hangul input device that ranks candidate syllable sequences corresponding to an undetermined button sequence inputted from the reduced Korean keyboard unit using a syllable context probability model.

The condensed Hangul keyboard part has 10 high frequency consonants 'ㄱ', 'ㄴ', 'ㄷ', 'ㄹ', 'ㅁ', 'ㅂ', 'ㅅ' and 'ㅇ'. 'ㅈ'. Assign 'ㅎ' to different buttons, and the other consonants 'ㄲ' for 'ㄱ', 'ㄸ' for 'ㄷ', 'ㅃ' for 'ㅂ', 'ㅆ' for 'ㅅ', 'ㅉ' 'J' and 'ㅊ' are 'ㅇ', 'ㅋ' is 'ㄹ' and 'ㅌ' is 'b'. 'Ct' is characterized by assigning to a button such as 'ㅁ'.

Another embodiment of the reduced Hangul keyboard part is limited to the representative vowels 'ㅏ', 'ㅓ', 'ㅗ', 'TT', 'ㅡ', 'ㅣ', and the representative vowels are 'ㄱ', It does not share buttons such as 'ㄴ', 'ㄹ' and 'ㅇ'.

Another embodiment of the reduced Hangul keyboard portion is (a) the phoneme 'a', 'ㄲ' is assigned to the button '1', (b) the phoneme 'b', 'ㅌ' is assigned to the button '2', (C) Phonemes 'ㄷ', 'ㄸ', 'ㅓ', 'ㅕ' are assigned to button '3', (d) Phonemes 'ㄹ', 'ㅋ' are assigned to button '4', ) Phonemes' ㅁ ',' '', 'ㅡ' are assigned to button '5', (f) Phonemes' ㅂ ',' ㅃ ',' ㅗ ',' ㅛ 'are assigned to button' 6 ', (G) Phonemes 'ㅅ', 'ㅆ', 'ㅜ'. 'ㅠ' is assigned to button '7', (h) phoneme 'ㅇ', 'ㅊ' is assigned to button '8', and (i) phoneme 'ㅈ', 'ㅉ', 'ㅣ' means 'button' 9 ', and (tea) the phonemes' ㅎ', 'ㅏ' and 'ㅑ' are assigned to the button '0'.

Another embodiment of the reduced Hangul keyboard portion is (a) the phoneme 'a', 'ㄲ' is assigned to the button '1', (b) the phoneme 'b', 'ㅌ' is assigned to the button '2', (C) Phonemes 'ㄷ', 'ㄸ', 'ㅓ', 'ㅕ' are assigned to button '3', (d) Phonemes 'ㄹ', 'ㅋ' are assigned to button '4', ) Phonemes' ㅁ ',' '', 'ㅡ' are assigned to button '5', (f) Phonemes' ㅂ ',' ㅃ ',' ㅏ ',' ㅑ 'are assigned to button' 6 ', (G) Phonemes 'ㅅ', 'ㅆ', 'ㅗ'. 'ㅛ' is assigned to button '7', (h) phoneme 'ㅇ', 'ㅊ' is assigned to button '8', and (i) phoneme 'ㅈ', 'ㅉ', 'ㅣ' means 'button' 9 ', and (tea) the phonemes' ㅎ', 'TT', and 'ㅠ' are assigned to the button '0'.

Another embodiment of the reduced Hangul keyboard portion is (a) the phoneme 'a', 'ㄲ' is assigned to the button '1', (b) the phoneme 'b', 'ㅌ' is assigned to the button '2', (C) Phonemes 'ㄷ', 'ㄸ', 'ㅏ', 'ㅑ' are assigned to button '3', (d) Phonemes 'ㄹ', 'ㅋ' are assigned to button '4', ) Phonemes' ㅁ ',' '', 'ㅡ' are assigned to button '5', (f) Phonemes' ㅂ ',' ㅃ ',' ㅗ ',' ㅛ 'are assigned to button' 6 ', (G) Phonemes 'ㅅ', 'ㅆ', 'ㅜ'. 'ㅠ' is assigned to button '7', (h) phoneme 'ㅇ', 'ㅊ' is assigned to button '8', and (i) phoneme 'ㅈ', 'ㅉ', 'ㅣ' means 'button' 9 ', and (tea) the phonemes' ㅎ', 'ㅓ' and 'ㅕ' are assigned to the button '0'.

Another embodiment of the reduced Hangul keyboard portion is (a) phoneme 'a', 'ㄲ', 'ㅑ' is assigned to the button '1', (b) phoneme 'b', 'ㅌ', 'ㅡ' is Is assigned to button '2', (c) the phonemes' ㄷ ',' ㄸ ',' ㅓ 'are assigned to button' 3 ', and (d) the letters' ㄹ', 'ㅋ', 'ㅛ' are buttons' 4 ', (e) Phonemes' ㅁ ',' '', 'ㅗ' are assigned to button '5', and (f) Phonemes 'ㅂ', 'ㅃ', 'ㅃ' are assigned to '6' And (i) the phonemes 'ㅅ', 'ㅆ' and 'ㅕ' are assigned to button '7', and (h) the phonemes 'ㅇ', 'ㅊ' and 'ㅠ' are assigned to button '8'. And (i) the phonemes 'ㅈ', 'ㅉ', and 'ㅣ' are assigned to the button '9', and (car) phonemes 'ㅎ' and 'TT' are assigned to the button '0'.

Another embodiment of the reduced Hangul keyboard portion is (a) the phoneme 'a', 'ㄲ', 'ㅕ' is assigned to the button '1', (b) the phoneme 'b', 'ㅌ', 'ㅑ' Is assigned to button '2', (c) the phonemes' ㄷ ',' ㄸ ',' ㅓ 'are assigned to button' 3 ', and (d) the letters' ㄹ', 'ㅋ', 'ㅛ' are buttons' 4 ', (e) Phonemes' ㅁ ',' '', 'ㅡ' are assigned to button '5', and (f) Phonemes 'ㅂ', 'ㅃ', 'ㅏ' are '6' And (i) the phonemes 'ㅅ', 'ㅆ' and 'ㅗ' are assigned to button '7', and (h) the phonemes 'ㅇ', 'ㅊ' and 'ㅠ' are assigned to button '8'. And (i) the phonemes 'ㅈ', 'ㅉ', and 'ㅣ' are assigned to the button '9', and (car) phonemes 'ㅎ' and 'TT' are assigned to the button '0'.

Another embodiment of the reduced Hangul keyboard portion is (a) the phoneme 'a', 'ㄲ', 'ㅕ' is assigned to the button '1', (b) the phoneme 'b', 'ㅌ', 'ㅑ' Is assigned to button '2', (c) the phonemes' ㄷ ',' ㄸ ',' ㅓ 'are assigned to button' 3 ', and (d) the letters' ㄹ', 'ㅋ', 'ㅛ' are buttons' 4 ', (e) Phonemes' ㅁ ',' '', 'TT' are assigned to button '5', and (f) Phonemes 'ㅂ', 'ㅃ', ',' are buttons '6' 'G', 'ㅆ', 'ㅡ' are assigned to button '7', and (h) the phonemes 'ㅇ', 'ㅊ' and 'ㅠ' are assigned to button '8' And (i) the phonemes 'ㅈ', 'ㅉ', and 'ㅏ' are assigned to the button '9', and (car) phonemes 'ㅎ' and 'ㅣ' are assigned to the button '0'.

Another embodiment of the reduced Hangul keyboard part is characterized in that all the phonemes are arranged as representative phonemes and the function buttons are removed in the above embodiments.

In addition, the non-deterministic Hangul input method using the syllable context probabilistic model and table-based parsing of the present invention has a function of pressing to input a phoneme button or a dependent phoneme in which one or more Hangul phonemes are arranged in a Jamo mixture on a scaled Hangul keyboard. A button for deciding the candidate currently output to the LCD display for the button, the undecided button sequence as a syllable sequence, a change button for outputting the next-order candidate to the LCD, a candidate selection button for moving to the candidate selection mode, and the most recently entered button Switching the operation mode to the Hangul input mode for Hangul input while allocating a delete button for canceling the phoneme button operation; Converting a code assigned to the button in response to receiving a button operation for selecting a button of the reduced Hangul keyboard; Receiving a phoneme button code from the reduced Korean keyboard unit, adding a button code to an undecided button sequence, generating candidate syllable sequences corresponding to the undecided button sequence, and determining a ranking; Receiving a function button code from the reduced Korean keyboard unit and setting or releasing the function button flag in a toggle method; Determining a syllable string by initializing a definite button code from the reduced Hangul keyboard and initializing data to be used for an undetermined button string and analysis; Receiving a change button code from the reduced Korean keyboard unit and setting a candidate selection mode flag if there are three or more candidates and replacing candidates displayed on a screen when two candidates are present; Receiving a delete button code from the reduced Korean keyboard unit and deleting the last element of the pending button sequence; Selecting and confirming a candidate to be outputted on the screen by receiving a button code from the reduced Korean keyboard unit with a candidate selection mode flag set; Presenting a list of candidate syllable sequences on a screen with the candidate selection mode flag set; And outputting the first-order candidate syllable sequence to a region corresponding to the indeterminate button sequence while the candidate selection mode flag is not set.

A table-based parsing technique is used to generate the candidate syllable sequence, and in order to reduce the amount of storage space, the analysis process maintains a single syllable cell, an intermediate result cell and a final result cell in the parsing table, and after the analysis is finished, Only the intermediate result cell and the end result cell are characterized.

The button string / syllable correspondence table may be used in the process of searching for a syllable corresponding to a button string to configure the single syllable cell.

The probability of occurrence of the candidate syllable sequence is calculated based on a syllable context probability model, and the probability is used as a criterion for ranking candidate syllable sequences.

In the process of extracting the syllable context probabilities, the syllable concatenation frequency considering spacing from the corpus and the syllable concatenation frequency without considering spacing are considered together.

In the process of extracting the syllable context probability, the utilization of the probability data is measured by using a scale that considers the syllable concatenation frequency, the rank of the syllable concatenation frequency, and the syllable context probability value itself. Is characterized in that to use only high probability data.

Compute a consonant distribution vector for each phoneme, calculate an ambiguity-resolved error occurrence estimator using a scale that is proportional to the magnitude of the individual vector for two vectors, and inversely proportional to the cosine of the vector, and estimate an ambiguity-resolved error occurrence. It is characterized by automatically searching for the optimum keyboard layout by selecting the minimum arrangement of the phonemes.

1 is a block diagram showing a preferred embodiment of a non-deterministic Hangul input device according to the invention,

Figure 2 is an illustration of a conventional Cheonjiin (trademark) keyboard arrangement.

Figure 3 is an illustration of a conventional handol (trade name) keyboard arrangement.

4 is an overall flow chart of the present invention.

5 is an exemplary view showing a first preferred embodiment of a reduced Hangul keyboard portion of the present invention.

6 is an exemplary diagram of a parsing table used to generate candidate syllable sequences of the present invention.

7 is a conceptual diagram illustrating the principle of expansion and contraction of the parsing table of the present invention.

8 is an exemplary view showing a part of a button string / syllable correspondence table of the present invention.

9 is a flowchart showing a preferred embodiment of the phoneme button processing unit of the present invention.

10 is an exemplary diagram showing a change of a parsing table in the process of inputting a button string "8339";

Fig. 11 is a flowchart showing a preferred embodiment of the function button processor of the present invention.

Fig. 12 is an illustration showing the configuration of a parsing table in a state in which button string "1739 # 901" is entered.

13 is a flowchart showing a preferred embodiment of the button processing unit of the present invention.

Fig. 14 is a flowchart showing a preferred embodiment of the change button processor of the present invention.

Fig. 15 is an illustration showing a configuration of a parsing table in a state in which button string " 10896002 " is entered.

Fig. 16 is an illustration showing a configuration of a parsing table in a state in which button string " 209839 " is entered.

Fig. 17 is a flowchart showing a preferred embodiment of the delete button processor of the present invention.

Fig. 18 is an illustration showing the configuration of a parsing table in a state in which button string "70405" is input.

Fig. 19 is a flowchart showing a preferred embodiment of the candidate selection processing section of the present invention.

20 is a view showing a preferred embodiment of the candidate output screen configuration and the screen change by the direction button.

21 is a flowchart showing a preferred embodiment of the screen output unit of the present invention.

Figure 22 is an exemplary view showing a second embodiment of the reduced Hangul keyboard portion of the present invention.

Figure 23 is an exemplary view showing a third embodiment of the reduced Hangul keyboard portion of the present invention.

24 is an exemplary view showing a fourth embodiment of a reduced Hangul keyboard portion of the present invention.

25 is an exemplary view showing a fifth embodiment of a reduced Hangul keyboard portion of the present invention.

26 is an exemplary view showing a sixth preferred embodiment of a reduced Hangul keyboard portion of the present invention.

<Explanation of symbols for the main parts of the drawings>

10: reduced type Hangul keyboard portion 20: button string parsing portion

21: phone button processing unit 22: function button processing unit

23: confirmation button processing unit 24: change button processing unit

25: delete button processing unit 30: button string / syllable correspondence table

40: syllable context probability data 50: parsing table

60: candidate selection processing unit 70: screen output unit

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram showing a preferred embodiment of a non-deterministic Hangul input device according to the present invention.

According to a preferred embodiment of the present invention, as shown in FIG. 1, a phoneme button in which one or more Hangul phonemes are arranged in a Jamo mixture, a function button pressed to input a dependent phoneme, and a candidate currently output to the LCD device. Confirming button for confirming that syllable string is syllable string, Change button for outputting next-order candidate to LCD, Candidate selection button for moving to candidate selection mode, and Delete button for canceling the most recent phoneme button operation. A reduced Hangul keyboard unit 10 for generating a code assigned to the button as a button manipulation occurs in the keyboard layout including a; A button string parsing unit (20) for processing the buttons input from the reduced Hangul keyboard (10) in the phoneme input mode to generate possible candidates and ranking of candidates; A button string / syllable correspondence table 30 which is a material used by the button string parser 20 to efficiently search for a syllable corresponding to a partial button string; Syllable context probability data (40) used in the button string parsing unit (20) to probabilistically rank the candidates; A parsing table (50) for storing candidate syllable sequences generated by the button sequence parsing section (20) and for recalculating when there is additional button input; A candidate selection processing unit 60 for determining a syllable sequence or changing a page of a candidate list to be output to the liquid crystal display according to a button input from the reduced type Hangul keyboard unit 10 while a candidate selection mode flag is set; It is configured to include a screen output unit 70 for outputting the appropriate contents to the liquid crystal display with reference to the parsing table 50.

In the button string parsing unit 20, the phoneme button processing unit 21, the function button processing unit 22, the confirmation button processing unit 23, the change button processing unit 24, and the delete button processing unit 25 according to the type of the input button. Is responsible for handling input buttons.

4 is a flowchart schematically showing a preferred embodiment of the reduced Hangul keyboard 10, the button sequence parser 20, the candidate selector 60 and the screen output unit 70 in the Hangul input method according to the present invention to be.

In the device to which the present invention is applied, a mode for inputting Hangul by the user is switched to the operation mode into the Hangul input mode (S10), and the reduced type Korean keyboard portion 10 converts the buttons input by the user into codes ( S20). The button sequence parsing unit 20 performs the candidate selection processing unit 60 when the candidate selection mode flag for selecting one of the multiple candidates is present (S30). The phoneme button processor 21, the function button processor 22, the confirmation button processor 23, the change button processor 24, and the delete button processor 25 are respectively executed according to the type of the input button (S50 to S140). . On the other hand, when a direction button used to move the cursor up, down, left, or right, or an input mode switch button used to switch the input mode to Korean, English, numbers, symbols, etc. is inputted, the undetermined button string is output to the current liquid crystal display device. After determining the syllable string (S150), the input button is processed by the common button processor (S160). Since the common button processing unit processing the direction button or the input mode switching button is outside the scope of the present invention, description thereof will be omitted. When the button string parsing unit 20 and the candidate selection processing unit 60 process the button input, the display output unit 70 displays the candidate syllable sequence in the page unit according to the rank if the current state is the candidate selection state. If not, the candidate syllable string whose rank corresponds to the output candidate number is output to the area corresponding to the undecided button string in the LCD (S180). In this case, the syllable sequence corresponding to the indeterminate button sequence may increase the user's recognition by changing the color or density of the letters or by classifying the syllable sequence using a specific symbol.

Prior to the description of the detailed flow of the Hangul input method according to the present invention. A preferred embodiment of a reduced Hangul keyboard according to the present invention is presented, and several terms and data structures are defined and explained.

Figure 5 is an exemplary view showing a preferred embodiment of a reduced Hangul keyboard portion according to the present invention.

In FIG. 5, the Hangul phoneme expressed in bold and large letters indicates a representative phoneme of the corresponding button, and the Hangul phoneme expressed in small and thin letters means a dependent phoneme. In FIG. 5, the button labeled Change is a change button, and the button labeled Function is a function button. Among the Korean consonants, the ten most frequent occurrences are 'ㄱ', 'ㄴ', 'ㄷ', 'ㄹ' and 'ㅁ'. '자', 'ㅅ', 'ㅇ', 'ㅈ' and 'ㅎ' and 6 phonemes with high frequency among Korean vowels' ㅏ ',' ㅓ ',' ㅗ ',' ㅜ ',' ㅡ ',' ㅣ 'is assigned as the representative letter. All consonants except the representative consonants are defined as dependent consonants, which are input by pressing the function button after pressing the function button. Among the vowels except the representative vowels, 'ㅑ', 'ㅕ', 'ㅛ', and 'ㅠ' are defined as subordinate vowels, and these phonemes are entered as sequential presses of function buttons and corresponding buttons as in subordinate consonants. . However, complex vowels such as 'ㅘ', 'ㅞ', 'ㅢ', etc. are configured to be input as a combination of representative vowels.

(Term Definition)

Reduced Hangul Keyboard: A Hangul keyboard with fewer buttons that make up the keyboard than the number of Hangul characters. In miniature Korean keyboards, to represent all kinds of phonemes, you typically define the function of a button to assign multiple phonemes to a button or to represent a phoneme through a combination of two or more buttons.

Representative Phoneme: When two or more phonemes are assigned to a button, it is the phoneme that is entered when the button is pressed alone. If the representative phoneme is a consonant, it is called a representative consonant.

Dependent phoneme: means any other phone except the representative phoneme. At least one button must be pressed to enter dependent phoneme. If the dependent phoneme is a consonant, it is called a dependent consonant.

Multi-type method: Pressing a button multiple times to enter the dependent phonemes assigned to that button.

Function button method: A method of inputting a dependent phoneme by pressing one or more function buttons before or after pressing a button to which the dependent phoneme is assigned.

Jamo Placement Detachable keyboard: A keyboard that has no consonants and vowels assigned to one button at the same time.

Jamo arrangement Mixed keyboard: A type of keyboard that can be assigned consonants and vowels simultaneously to a single button. In a consonant arrangement mixed keyboard, one consonant and one vowel may be assigned as representative phonemes for one button, and thus a button having a plurality of representative phonemes may exist.

Ambiguity: In a reduced Hangul keyboard, more than one phoneme can be assigned to a button, so when a user presses a button, it is crucial to determine which phoneme the user intends to input among the phonemes assigned to that button. A situation may arise that cannot be done. Similarly, when a plurality of button sequences are consecutively pressed, there is a situation where it is impossible to determine which of the corresponding syllable sequences is the syllable sequence that the user wants to input. This phenomenon is called ambiguity.

Deterministic phoneme input method: A method of inputting a phoneme so that ambiguity does not occur by defining a keyboard such that only one phoneme corresponding to one button or button string exists. In the deterministic phoneme input method, the syllable string corresponding to the button string pressed by the user when entering Hangul is unique, and the syllable string corresponding to the input intention is input unless the user makes an error.

Nondeterministic Phoneme Input Method: A phoneme input method that allows a case where there are multiple phonemes corresponding to one button or button string. In the non-deterministic phoneme input method, there are several syllable strings that correspond to the button strings pressed by the user, so that ambiguity occurs. The process is necessary.

Indeterminate button sequence: In the case of nondeterministic phoneme input method, ambiguity occurs, and the entire ambiguity cannot be solved automatically, so the user has to confirm or correct the input by additional button manipulation. At this time, the button sequence that has not been confirmed by the user is called an indeterminate button sequence. When the button is pressed in the nondeterministic phoneme input method, the change occurs only in the syllable sequence corresponding to the indeterminate button sequence of the entire input text.

Candidate syllable sequence: In the non-deterministic phoneme input method, there are a large number of syllable sequences corresponding to the undetermined button sequences, each of which is called a candidate syllable sequence. In the present invention, candidate syllable sequences may be abbreviated as candidates.

Candidate Ranking Criteria: In the non-deterministic phoneme input method, if there are a plurality of candidate syllable sequences, their ranks are determined and the candidate syllable sequences corresponding to the first rank are output to the liquid crystal display. To this end, a criterion for determining the rank of candidate syllable sequences should be provided, which is called a candidate ranking criterion.

Syllable unit ambiguity resolution method: A method of resolving ambiguity for each syllable input among nondeterministic phoneme input methods. In the syllable unit ambiguity resolution method, only syllables composed of one syllable among the syllable sequences corresponding to the undecided button sequence are included in the candidate syllable sequence. In this method, when a plurality of candidate syllable strings exist when the input for one syllable is completed, the user is required to manipulate the ambiguity.

Word unit ambiguity resolution method: A method of resolving ambiguity for each word input among nondeterministic phoneme input methods. In the word unit ambiguity resolution method, only the syllable sequences included in the word list among the syllable sequences corresponding to the undecided button sequence are included in the candidate syllable sequence. In this method, if an error occurs in the process of resolving ambiguity when the input of one word is finished, the user is required to manipulate the ambiguity.

Arbitrary unit ambiguity resolution method: A method in which a user can arbitrarily determine a unit that resolves ambiguity among non-deterministic phoneme input methods. That is, the user can perform an operation for ambiguity resolution at any time. In random unit ambiguity resolution, candidate syllable sequences generally do not exceed one word unit. The reason for this is that in a reduced Hangul keyboard, a space character is generally assigned to a button different from the Hangul phoneme, so that the syllables preceding the space character in the first-order candidate syllable string are not changed by subsequent button manipulations and thus are processed decisively.

Same-button ambiguity: In a multi-type method, when you enter a syllable sequence in which the last syllable's last syllable and the next syllable's initial are assigned to the same button, you can determine whether the button pressed at the syllable boundary belongs to the previous syllable or the next syllable. A number of things happen, called equal button ambiguity. In general, in multi-type method, in order to solve the same button ambiguity, a syllable boundary is distinguished by pressing a function button indicating a syllable boundary or by not pressing the button for a certain period of time. It acts as a dropping factor.

Consonant ambiguity: When you press a button that has consonants and vowels assigned in a consonant layout mixed keyboard, you can't tell whether the phoneme you want to input is consonant or vowel. This is called consonant ambiguity. Deterministic phoneme input methods generally do not use the Jamo batch mixed keyboard because the Jamo-division ambiguity cannot be completely eliminated without additional button manipulation.

Syllable boundary ambiguity: In the nondeterministic phoneme input method, the ambiguity caused by not being able to boundary the pressed button sequence by syllable unit is called syllable boundary ambiguity.

Ambiguity Resolving Rate: A ratio of automatically eliminating ambiguity for the ambiguity generated by the non-deterministic phoneme input method without requiring additional user button manipulation. In other words, the ambiguity resolution rate is the rate at which the first-order candidate string matches the string that the user wants to input.

Hangul Discipleship Principle: In syllables, syllables consist of consonants + neutral or consonants + neutral + jongseong, which consists of one consonant, one or more vowels, and one or more consonants. It means the principle that the possible consonants are limited.

Complete syllable: Korean syllables belonging to KSC5601 character set. sure ??, ?? Although there are syllables such as Hangul syllables and not belonging to the KSC5061 character set, most devices that use the reduced Hangul keyboard determine whether or not they are Hangul based on the KSC5601 character set.

Impossible syllable: " "And" Like ", it means Hangul syllable that does not contradict the Hangul disciple principle but does not correspond to a complete syllable.

Incomplete syllables: It means syllables that consist of Hangul phonemes or phoneme strings but are contrary to the principles of Hangul discipleship. Incomplete syllables are syllables that consist only of consonants or syllables without consonants.

Primary syllables: Incomplete syllables consisting of only the first consonants.

Neutral Syllable: An incomplete syllable consisting of only neutral.

Table-based parsing: Analyze the structure of an input token sequence using a triangular table. Table-based parsing saves intermediate calculations in the parsing table, so even if tokens are added, the result of analysis can be obtained by parsing only the parts related to the newly added token without re-parsing the entire input token sequence. Thus, table-based parsing has the feature of eliminating duplicate calculations. Since table-based parsing is widely known in the conventional natural language processing research, a detailed description thereof will be omitted. In the present invention, table-based parsing is applied by considering an undetermined button sequence as an input token sequence.

Parsing Table: A data structure used in table-based parsing, as shown in FIG. The number of rows and columns in the parsing table is equal to the length of the input token column. In the present invention, the cells in the i th row and the j th column of the parsing table are expressed as C (i, j). Cell C (i, j) contains the parsing result of the partial token sequence consisting of the i th to j th of the input token sequences, and the contents are constructed when the length of the pending button sequence is j, and thereafter, it does not change.

Parsing table size: The number of columns or rows that make up a parsing table. The size of the parsing table is equal to the length of the input token string.

Expansion of the parsing table: As the length of the input token, i.e. the undetermined button row, increases, the size of the parsing table must increase, which is called the expansion of the parsing table. When the phoneme button is input and the pending button column is increased, as shown in FIG. 7, the number of columns and rows of the parsing table is increased by one. At this time, the extended part of the parsing table is the last column and row.

Reduction of parsing table: If the length of the pending button column is reduced because the delete button is pressed, the parsing table is reduced in size by removing the last column and row as shown in FIG.

Partial button sequence: Refers to a part of the pending button sequence, and represents a partial button sequence consisting of the i th to j th buttons among the undecided button sequence as I (i: j).

Single-syllable cell: When the length of the undetermined button sequence is n, the cells corresponding to all C (i, n) in the parsing table are called single-syllable cells. In the single syllable cell C (i, n), single syllables corresponding to the partial button string I (i: n) may be candidates. Single syllables that can be included in a single syllable cell can be found by querying a button sequence / syllable correspondence table for syllables corresponding to the partial button sequence I (i: n). FIG. 8 shows a part of a button string / syllable correspondence table that can be used when adopting the reduced Hangul keyboard shown in FIG. The button sequence / syllable correspondence table contains the button sequence, the corresponding syllable, the probability that the syllable will appear after the non-Hangul characters, and the probability that the syllable will appear after the Hangul characters. Hangul syllables in the button / syllable correspondence table include complete syllables and consonant single syllables. The circled characters in the button column of the button string / syllable correspondence table indicate that the button has a function button bit set to be described later in the corresponding button code.

Intermediate result cell: As shown in FIG. 6, a cell corresponding to all C (1, j) in the parsing table is referred to as an intermediate result cell. The intermediate result cell C (1, j) contains candidate syllable strings corresponding to the partial button strings I (1: j).

Final result cell: When the length of the undetermined button row is n, the column C (1, n) is called the final result cell. Since the final result cell contains candidates for the entire undecided button sequence, the screen output unit outputs the screen using the candidates in the final result cell.

Candidate Selection Mode Flag: In the present invention, since the non-deterministic phoneme input method is adopted, the number of syllable strings corresponding to the indeterminate button sequence may be many. If the first-priority candidate is not the input intended by the user, the candidate selection mode may be switched to select a desired candidate while the candidate lists are displayed on the screen, and the candidate selection mode flag indicates whether the current state is the candidate selection mode. Boolean variable.

Current candidate page: In the candidate selection mode, a variable indicating how many pages the candidate list shown on the current screen corresponds to.

Candidate number: means the number of candidate syllable strings included in the final result cell.

Function button flag: A variable that toggles as the function button is pressed. When the phoneme button is pressed while the function button flag is set, the function button bit is additionally set in the code of the button and added to the pending button string. When the phoneme button is pressed, the function button flag is reset.

Current candidate number: Refers to a candidate number of syllable sequences that is output to the liquid crystal display in correspondence with the indeterminate button sequence. Generally, the first-order candidate is output to the liquid crystal display, but if the change button is pressed when the number of candidates is two, the rank of the candidate not output to the liquid crystal display is set to the current candidate number.

9 is a flowchart illustrating a preferred embodiment of the phoneme button processing unit 21 in the Hangul input method according to the present invention. When the phoneme button processing unit 21 is called, it is checked whether the current function button flag is set (S610). If the function button flag is set, the function button bit is set in the phoneme button code and added to the pending button string (S620). The phoneme button code is added to the pending button string as it is (S630). When the phoneme button code is added to the pending button row, the size of the parsing table 50 is expanded (S640) and a candidate is added to single syllable cells in the expanded cell area (S650). The operation of adding a candidate to a single syllable cell is performed by registering a syllable as a candidate when there is a corresponding syllable for a button sequence that the single syllable cell means by referring to a button string / syllable correspondence table. After the configuration of the extended single syllable cell is completed, the final result cell is performed by combining all the candidates in the intermediate result cells and the extended single syllable cell (S660). When the length of the pending button row is n, the final result cell C (1, n) includes the intermediate result cell C (1, i) and the single syllable cell C (i + 1,) for all i between 1 and (n-1). The combination of candidates in n) may be added. At this time, candidates whose last syllables are incomplete syllables among candidates in the intermediate result cell are excluded from the combination consideration. In addition, when there is at least one candidate in which the last syllable is a complete syllable in the final result cell, the candidate in which the last syllable is an incomplete syllable is removed. If candidates for the final result cell are determined, the candidates are ranked according to candidate ranking criteria (S670). Thereafter, the operation goes to step S180 of FIG. 4.

10 is a button '8', '3', '3' in the reduced Hangul keyboard portion 10 designed in the present invention. When '9' is sequentially pressed, the changing process of the parsing table 50 according to the processing of the phoneme button processing unit 21 is shown. In FIG. 10, numbers written next to the parsing table 50 indicate buttons corresponding to the corresponding rows and columns, and each cell includes a rank of the candidate syllable sequence, a candidate syllable sequence, and a probability value of the candidate syllable sequence.

Initially, the size of the parsing table 50 starts with 0, and when the first button '8' is pressed, the parsing table T101 is generated. According to the button string / syllable correspondence table 30 shown in FIG. 8 in the process of constructing a single syllable cell, since the partial button string "8" corresponds to the syllable "o" in the button string / syllable correspondence table, the cell C (1,1) is " O is added as a candidate. As a result, the screen output unit 70 outputs the first rank candidate syllable string "o" of the final result cell C (1,1) to the liquid crystal display.

When the second button '3' is pressed, the parsing table T102 is created. In the process of constructing a syllable cell, the partial button string "3" corresponds to the syllable "c", so "c" is added to cell C (2,2) as a candidate, and the partial button string "83" is assigned to the syllable "er". Correspondingly, "uh" is added as a candidate to cell C (1,2). Meanwhile, in the step of combining the intermediate result cells and the single syllable cells, the final result cell C (1,2) may be generated as a combination of the intermediate result cell C (1,1) and the single syllable cell C (2,2). Since candidates in the monosyllable cells C (1,2) are all composed of incomplete syllables, no candidates are added. As a result, the screen output unit 70 outputs, to the liquid crystal display, "er" which is the first-order candidate syllable sequence of the final result cell C (2,2).

When the third button '3' is pressed, a parsing table T103 is created. In the process of constructing the syllable cell, the partial button sequence "3" corresponds to the syllable "c", the partial button sequence "33" corresponds to the syllable "more", and the partial button sequence "833" corresponds to the syllable "get". "C" is added to (3,3), "more" is added to single-syllable cell C (2,3), and "get" is added to single-syllable cell C (1,3), respectively. In the process of constructing the final result cell by the combination of cells, a combination of cell C (1,1) and cell C (2,3) and cell C (1,2) and cell are included in the final result cell C (1,3). Candidates can be added to the combination of C (3,3), but in the former combination all candidates in the intermediate result cell C (1,1) are removed because they are incomplete syllables, and in the latter combination, the " Because there is a candidate consisting only of "complete syllables", the syllable sequence "c" that is the result of the combination is not added as a candidate. As a result, the screen output unit 70 outputs, to the liquid crystal display device, "get", which is the first-order candidate syllable sequence of the final result cell C (3,3).

When the fourth button '9' is pressed, parsing table T104 is created. In the process of constructing a syllable cell, the partial button sequence "9" corresponds to the syllable "ㅈ", the partial button sequence "39" corresponds to the syllable "D", and the partial button sequence "339" corresponds to the syllable "De". "J" is added to (4,4), "D" is added to single-syllable cell C (3,4), and "de" is added to single-syllable cell C (2,4) as a candidate. Meanwhile, in the process of constructing the final result cell, the cell C (1,1), the cell C (2,4), the cell C (1,2), the cell C (3,4), the cell C (1,3) and You can consider combining cells C (4,4), where the first combination cannot be combined because all candidates in the intermediate result cell C (1,1) end with incomplete syllables. Because it satisfies, the syllable sequence "where" is generated and added as a candidate to the final result cell C (4,4). In the third combination, since there is already a candidate consisting of only complete syllables in the final result cell C (4,4), the syllable sequence "get" produced as a result of the combination is not added as a candidate. As a result, the screen output unit 70 outputs "where", which is the first order candidate syllable sequence of the final result cell C (4,4), to the liquid crystal display.

As can be seen from the above example, each time the phoneme button is pressed, a single syllable cell is configured and a final result cell is constructed. Since the maximum button string length that can constitute one syllable is 5, the syllable is a single syllable. When configuring a cell, it is necessary to refer to a table of up to five button string / syllable correspondence tables, and the calculation complexity is very low because only five combinations of cells need to be performed when configuring the final result cell. On the other hand, in the case of using basic table-based parsing, n (n + 1) / 2 cells should be maintained when the length of the indeterminate button column is n. However, in the present invention, the table-based parsing is partially modified to configure the final result cell. After finishing, remove the remaining cells except the intermediate result cell and the final result cell to keep only the same number of cells as the length of the pending button row. For example, in FIG. 10, after the configuration of the table T102, only the cells C (1,1) and C (1,2), and after the configuration of the T103 is completed, the cells C (1,1) and C (1, 2) and C (1,3) only, after the configuration of T104, only cells C (1,1), C (1,2), C (1,3) and C (1,4) remain and the rest The cells are removed. In the exemplary drawings presented in the present invention, cells removed when the parsing table is represented are also represented, which is only to help understand the intermediate process. The modified table parsing produces the same results as the existing table parsing for the present invention because the cells to be removed are not used in subsequent calculations. According to the modified table parsing, when the length of the undetermined button row is n, at most (n + 5) in the parsing process and only n cells after parsing are required, the present invention provides a method for storing the required storage location. It is very small in size.

11 is a flowchart illustrating a preferred embodiment of the function button processing unit 22 in the Hangul input method according to the present invention. The function button processing unit 22 basically serves to change the function button flag in a toggle manner. When the function button processing unit 22 is called, it is checked whether the current function button flag is set (S810). If the flag is already set, the flag is cleared (S820). Otherwise, the flag is set (S830).

12 is '1' in the reduced Hangul keyboard portion 10 created in the present invention. The configuration of the parsing table 50 is shown when '7', '3', '9', '#', '9', '0' and '1' are pressed in sequence. If the fifth button '#' is pressed, the function button flag is set according to the processing of the step S830 of FIG. 11, and if the sixth button '9' is pressed, the processing of the step S620 of FIG. Accordingly, '⑨' with the function button bit set to button '9' is added to the undecided button sequence.

13 is a flowchart showing a preferred embodiment of the button processing unit 23 in the Hangul input method according to the present invention in detail. When the confirmation button is pressed, the confirmation button processing unit 23 performs the syllable sequence determination operation (S1010). In the syllable sequence, the function button flag is cleared, the parsing table is initialized, and the pending button sequence is initialized. In the present invention, the confirmation operation of the syllable sequence is performed when the confirmation button is pressed (S1010 portion in FIG. 13), when a mode switch button or direction button is pressed during Hangul input (in S150 portion in FIG. 4), and in the candidate selection mode to be described later. Occurs when a candidate is selected (part S420 in FIG. 19). If you decide the syllable sequence by pressing the confirm button among these, the length of the word to input is very long, so it is divided and input. If the word to be input is long, the number of candidate syllable sequences that can be generated increases, making it difficult to resolve the ambiguity. Therefore, in the present invention, by allowing the words to be divided by using the confirm button, the syllable sequence to be input does not exist in the first rank, thereby reducing the inconvenience of the candidate selection process.

14 is a flowchart illustrating a preferred embodiment of the change button processing unit 23 in the Hangul input method according to the present invention. When the change button is input, it is checked whether the number of candidates is three or more (S1210). If the number of candidates is three or more, the candidate selection mode flag is set and the current candidate page is set to 1 (S1220). Otherwise, the current candidate number indicating the rank of the candidate currently output on the LCD is equal to the number of candidates. (S1230). If this condition is met, the current candidate number is set to 1 (S1240), otherwise, the current candidate number is increased by 1 (S1250).

An example of a process of processing a change button when there are two candidates in the final result cell will be described with reference to FIG. 15. FIG. 15 is a parsing table generated when a button is pressed in order to input a syllable string "highlighted" in the reduced Hangul keyboard unit 10 of the present invention. According to FIG. 15 when the buttons '1', '0', '8', '9', '6', '0', '0', and '2' are sequentially pressed to input syllable strings "highlight" There are two candidate syllable sequences "joined" and "highlighted" in the final result cell. However, since the ranking candidate syllable sequence is "subscribed", the syllable sequence "subscribed" appears in the liquid crystal display instead of the user intended. In this case, when the user presses the candidate change button '*' to change the candidate, since the number of candidates currently in the final result cell is two, the current candidate number is set to 2 by the process of step S1250 of FIG. 14. Through the processing of the screen output unit 70, " highlighted " the second-order candidate is output on the screen.

An example of a process of processing a change button when there are three or more candidates in the final result cell will be described with reference to FIG. 16. FIG. 16 is a parsing table generated when a button is pressed in order to input a syllable string of "daytime" in the reduced-size Korean keyboard unit 10 of the present invention. According to FIG. 16, when the buttons '2', '0', '9', '8', '3', and '9' are sequentially pressed to input the syllable sequence "daytime", five candidates are displayed in the final result cell. There are syllable strings "in", "in the daytime", "inner", "coldness" and "daytime". However, since the first-order candidate syllable sequence is "in", the syllable sequence of "in" appears instead of "in the day" intended by the user. In this case, when the user presses the candidate change button '*' to change the candidate, the candidate selection mode flag is set by the process of step S1220 of FIG. 14 because the number of candidates currently in the final result cell is five. Through the processing of the screen output unit 70, which will be described later, five candidates are presented on the screen to select the intended "day."

17 is a flowchart showing a preferred embodiment of the delete button processing unit 23 in the Hangul input method according to the present invention. When the delete button is input, it is checked whether the length of the pending button row is 0 (S1410). If the length of the pending button row is 0, it means that there are no more elements to delete, so the process ends. Otherwise, the last button of the pending button row is removed (S1420), the size of the parsing table is reduced, and the current candidate number is set to 1 ( S1430).

18 illustrates a state (T181) of the parsing table when the buttons '7', '0', '4', '0', and '5' are sequentially pressed in the reduced-size Korean keyboard unit 10 of the present invention. This shows the state of the parsing table (T182) when the delete button is pressed. When the buttons '7', '0', '4', '0', and '5' are pressed, the first person candidate syllable sequence of the final result cell C (5,5) of the parsing table T181 is liquid crystal. It is output to the display device. In this state, if the delete button is pressed, the length of the pending button row is reduced by one, and the parsing table is reduced to become T182. As a result, "Person", which is the first order candidate syllable sequence of the final result cell C (4,4), is output to the liquid crystal display. In this process, the parsing table is only reduced and no recalculation occurs, so the delete button is very fast.

19 is a flowchart illustrating a preferred embodiment of the candidate selection processing unit 23 in the Hangul input method according to the present invention. The candidate selection processor 23 first checks whether the input button corresponds to a number (S410). If this condition is correct, the current candidate number is set to the corresponding number, the candidate selection mode flag is released, and the syllable confirmation operation is performed. If the condition checked in step S410 is not met, it is checked whether the input button is a delete button (S430). If the input button is the delete button, the candidate selection mode flag is released and then the process moves to step S140 of executing the delete button processing unit 23. Otherwise, it is checked whether the input button is a direction button (S440). If the input button is a direction button, the candidate page output to the liquid crystal display is changed (S450).

FIG. 20 is a view illustrating a case in which a state of a parsing table is the same as that of FIG. 16 and a candidate is output from a device having a maximum number of candidates that can be represented on one screen (L201 and L202) and the direction buttons are output. In the beginning of the conversion to the candidate selection mode, since the current candidate page is 1, L201 of FIG. 20 is presented on the screen. Subsequently, when the right button is pressed, the number of candidate pages is increased by the S450 portion of FIG. 19, and L202 of FIG. 20 is presented on the screen by the screen output unit 70. Similarly, if the left button is pressed in this state, the number of candidate pages is decreased by step S450 of FIG. 19, and L201 of FIG. 20 is presented on the screen. Then, when the user presses the number button '2' while L201 of FIG. 20 is displayed on the screen, the candidate selection mode flag is released, and the second candidate "day" is output to a portion corresponding to the pending button row of the LCD. .

21 is a detailed flowchart illustrating a preferred embodiment of the screen output unit 70 in the Hangul input method according to the present invention. The screen output unit 70 checks whether the candidate selection mode flag is set (S1810). If this flag is set, the list of candidate syllable strings is output to the LCD of the screen as shown in L201 and L202 of FIG. 20 (S1820). Otherwise, the first candidate candidate syllable string is output to the region corresponding to the undecided button string (S1830).

Meanwhile, in the present invention, when there are several candidate syllable sequences in one cell, candidate ranking criteria for determining the ranking between them are required. In the present invention, a syllable context probability model based on a probability of generating a corresponding syllable sequence is used as a candidate ranking criterion. In a syllable context probability model, Being syllable Given is the probability that the syllable will occur The formula for calculating is as shown in Equation 1 below. In Equation 1 'Is a symbol representing a syllable string boundary.

However, since it is difficult to obtain all the probability information used in Equation 1, in the conventional natural language processing research, the preceding syllable for each syllable Probability of a given syllable when two syllables precede it By using the approximation defined as in the following equation (2).

In the present invention, the probability of the syllable sequence is calculated based on Equation 2. Set to 1 to use. However, as the storage capacity of devices using the reduced Hangul keyboard increases in the future,It is obvious that the value can be increased to a larger value. Equation 3 is Syllable context probability model at

Inter-syllable concatenation probability required for syllable context probability model based on Equation 3 Is obtained using statistical information from large corpus. At this time There are two problems with using. One is that even if you use statistical information from large corpuses, there may be syllable pairs that do not appear in contiguous fashion. Probability in this case Becomes 0, and as a result, the probability of occurrence of syllable There is also a problem of being zero. This phenomenon is called data shortage problem in conventional natural language processing techniques. The other problem is Even if the value is minimized to 1, the number of probability data used in the probability model is 5,522,500 (= 2350 * 2350), which requires a large amount of storage.

The present invention solves the two problems mentioned above as follows. First, the data shortage problem is solved using the smoothing technique. Since the planarization technique is a technique widely used in the conventional natural language processing technique, the description thereof is omitted here. Next, in order to solve the problem of the size of the probability data, the number is reduced through the probability data screening process.

Since the screening of the probability data uses only a part of the entire probability data, it can lead to a loss of information, and the loss of the information can lead to a decrease in the ambiguity resolution rate. Therefore, in the present invention, the probability data is measured for individual probability data and the probability data with high utilization is preferentially selected. Although there are various measures for calculating the utilization of the probability data, in the present invention, individual syllable context probability With yourself, syllable And syllable Frequency of occurrences And all middle Define and use a measure that considers the ranking of a compound.

On the other hand, syllable context probability In the corpus Wow And then calculate using Equation 4.

However, in a mobile communication terminal in which a reduced Hangul keyboard is typically used, it is often inconvenient to insert a space character while inputting a Hangul syllable string. In consideration of this phenomenon, Is obtained using a modified equation as shown in Equation 5 below.

In equation (5) The syllable And syllable The frequency at which spaces or non-Hangul characters appear in between, Wow Is a constant that indicates how much space is taken into account. The sum is 1, and is an empirically determined number. this Wow May be appropriately changed according to the characteristics of the device to which the present invention is applied and the habit of the user who uses the device. For example, if a user thoroughly observes spacing on a device to which the present invention is applied. So that the value is close to 1. It would be reasonable to set the value to close to zero, and conversely if users ignore the spacing Value and It would be reasonable to set both values to 0.5.

In the nondeterministic phoneme input method, the degree of ambiguity and the difficulty of ambiguity vary depending on the order in which the phonemes are arranged on the keyboard. In addition, the ambiguity resolution rate is inversely proportional to the ambiguity generation amount and the ambiguity difficulty. Therefore, in the non-deterministic phoneme input method, it is important to find a keyboard arrangement that can increase the ambiguity resolution rate. To do this, simulations of all keyboard layouts can be used to select an empirical method of finding the keyboard layout with the highest ambiguity resolution. However, because there are too many ways to place 19 consonants and 10 vowels on 10 buttons, it is practically impossible to find an optimal keyboard layout through empirical methods. In addition, even if the keyboard layout can maximize the ambiguity resolution rate, if it seems cognitively difficult for the user, the product value is rather deteriorated. Accordingly, the present invention proposes a method for automatically finding a keyboard layout that can increase the ambiguity resolution rate without being cognitively difficult in the non-deterministic phoneme input method.

In the present invention, the optimum keyboard layout is generated in two steps. In the first step, the cognitive characteristics are considered, and the consonants or vowels are manually placed on the keyboard using the frequency of occurrence of phonemes, the lexical order of the phonemes, and the phonetic relationships between the consonants. In the second step, the rest of the letters are automatically placed on the keyboard to reduce the ambiguity and the difficulty of ambiguity by using the consonant distributions of the already placed letters and the rest of the letters.

The process of the first stage is as follows. First, the consonants or vowels are placed first on the keyboard, taking into account the cognitive characteristics of the keyboard. Considering the cognitive characteristics, the method of arranging consonants or vowels first can be considered to follow the dictionary order of the phonemes. 5 is an example of a case where consonants are first arranged in a lexical order. In the case of Hangul consonants, the number is 19, which is larger than the number of buttons. Therefore, 10 phonemes with a high frequency of appearance are first arranged as representative phonemes in a lexical order. On the same button as the similar representative phoneme, the phonemes are properly placed on the remaining buttons.

In the second step, the phoneme concatenation distribution from the corpus is used. The consonant concatenation distribution expresses the frequency of consonants that appear in concatenation before and after the consonant, in vector form, for each consonant. Ambiguity occurs in the nondeterministic phoneme input method because a plurality of representative or dependent phonemes can be assigned to one button. Thus, the amount of ambiguity is proportional to the frequency of the phonemes assigned to the buttons where the ambiguity occurs, which means that it is proportional to the size of the phoneme concatenation distribution vector of the phonemes assigned to the buttons where the ambiguity occurs. On the other hand, the difficulty of ambiguity means that the probability of occurrence of candidate syllable sequences calculated by Equation 3 is similar when ambiguity occurs. The probability of occurrence of a candidate syllable sequence is affected by the syllable context probability and the syllable context probability Since it is significantly related to the concatenation probability of, it can be seen that the probability of similarity of occurrences of candidate syllable sequences is proportional to the similarity of the consonant distribution of the phonemes assigned to the buttons where ambiguity occurs. Based on this principle, in the present invention, the phoneme concatenation distribution vector is , And the angle between the two vectors Pharyngeal phlegm , Ambiguity resolution error estimator Is defined as in Equation 6 below. Function in equation (6) Is , Increases when the monotonic increases, Is a monotonically decreasing function.

Given a keyboard layout based on Equation 6, the ambiguity error occurrence estimator of the corresponding keyboard layout corresponds to the sum of the values of Equation 6 calculated for all the buttons with ambiguity. On the other hand, it is not practical to calculate the ambiguity error occurrence estimator for all possible keyboard layouts because the number of unpositioned phonemes has been reduced by the first step of the keyboard layout determination. Accordingly, in the present invention, in addition to the keyboard arrangement of FIG. 5, the first stage of the keyboard arrangement and the function of Equation 6 22 to 26 show other embodiments of the reduced Hangul keyboard portion obtained by varying.

5, 22, and 23, the keyboard arrangements are arranged sequentially in consideration of cognitive factors, and then six vowels 'ㅏ', 'ㅓ', 'ㅗ', 'TT', Set 'ㅡ', 'ㅣ' as representative vowels, and the remaining four vowels 'ㅑ', 'ㅕ', 'ㅛ', and 'ㅠ' are 'ㅏ', 'ㅓ', 'ㅗ', 'TT' and It is placed as a dependent collection on the same button. At this time, to reduce the amount of ambiguity, six high frequency vowels are arranged exclusively with four high frequency consonants 'ㄱ', 'b', 'ㄹ' and 'ㅇ'. These three keyboard layouts are the functions used in equation (6). The difference occurs depending on what you choose.

24, 25, and 26, the arrangement of the consonants is sequentially arranged in consideration of cognitive factors, and then the entire ten vowels are arranged as representative vowels using Equation (6). These three keyboard layouts are the functions used in equation (6). The difference occurs depending on what you choose.

300KB syllable context probability data is applied by applying the above-described nondeterministic phoneme input method, syllable context probability model, table-based parsing, non-deterministic phoneme input method, selection of probability data, keyboard layout to minimize ambiguity. Table 2 shows the simulation results of Hangul input for 21,062,496 Korean words.

In FIG. 2, "removing FIG. 26 + function button" refers to a method of using the keyboard layout of FIG. 26, but removing the function button and recognizing all phonemes as representative phonemes. The "average number of button manipulations per syllable" in FIG. 2 is a result including the number of button manipulations required for properly inputting a word intended by a user, such as a candidate change button or a candidate confirmation button, when the ambiguity is unsuccessful. Based on the results in Table 2, all the methods using the function buttons show an ambiguity resolution of about 98%, so there is little user's cognitive burden. On the other hand, the method of FIG. 26 + function button removal has a merit that the number of button manipulations per syllable is small, but the ambiguity resolution rate is relatively low, so that the user may feel cognitively uncomfortable.

As can be seen in Table 2, the Hangul input method proposed in the present invention can input Hangul with only a similar number of button manipulations as the quality 2 beol type keyboard used in a personal computer. It can be seen that the Hangul input speed is faster than. In addition, the syllable context probability data used in this simulation is only 300KB, which is small enough to be applied to a mobile terminal or a device using another reduced Hangul keyboard.

Terminologies used in the present invention are terms defined in consideration of functions in the present invention, which may vary according to the intention or practice of those skilled in the art, and the definitions should be made based on the contents throughout the present application. something to do.

In addition, since the present invention has been described through the preferred embodiment of the present invention, in view of the technical difficulty of the present invention, a person having ordinary skill in the art may easily change other embodiments and other embodiments of the present invention. As can be added, it is apparent that all embodiments and modifications cited in the above description belong to the claims of the present invention.

As described in detail above, in developing a Hangul input device having a nondeterministic phoneme input method in a device using a reduced Hangul keyboard, a method for finding a syllable context probability model and a keyboard arrangement with less ambiguity invented in the present invention. Can be used to increase the ambiguity resolution rate, and table-based parsing speeds up the process by eliminating the recalculation according to the input and deletion of the phoneme. Therefore, there are various outstanding effects, such as being able to apply a practically nondeterministic phoneme input method to a Hangul input using a reduced Hangul keyboard while maintaining the reduced number of button manipulations.

Claims (17)

Phoneme button in which one or more Hangul phonemes are arranged in a Jamo mixture, function button pressed to input dependent phonemes, confirm button confirming that the candidate syllable string currently output on the liquid crystal display is the syllable string to be input, the next candidate A reduced Hangul keyboard portion including a change button for outputting a to a liquid crystal display, a candidate select button for moving to a candidate selection mode, and a delete button for canceling a most recently inputted phoneme button operation; And And a candidate syllable sequence corresponding to an undetermined button sequence inputted from the reduced Korean keyboard unit is ranked using a syllable context probability model. The Hangul input device of claim 1, wherein all the phonemes are assigned as representative phonemes and a modification is made to remove the function button. According to claim 1 or 2, wherein the reduced Hangul keyboard portion Ten high-frequency consonants 'ㄱ', 'ㄴ', 'ㄷ', 'ㄹ', 'ㅁ', 'ㅂ', 'ㅅ', 'ㅇ'. 'ㅈ'. Assign 'ㅎ' to different buttons, and the remaining consonants' ㄲ 'for' ㄱ ',' ㄸ 'for' ㄷ ',' ㅃ 'for' ㅂ ',' ㅆ 'for' ㅅ ',' 'ㅉ' for 'ㅈ', 'ㅊ' for 'ㅇ', 'ㅋ' for 'ㄹ' and 'ㅌ' for 'ㄴ'. 'T' is a Korean input device, characterized in that assigned to the button such as 'ㅁ'. According to claim 1 or 2, wherein the reduced Hangul keyboard portion, Representative vowels are limited to 6, 'ㅏ', 'ㅓ', 'ㅗ', 'TT', 'ㅡ', and 'ㅣ', and these representative collections are 'ㄱ', 'ㄴ', 'ㄹ', 'ㅇ' Hangul input device, characterized in that does not share the same button. According to claim 1 or 2, wherein the reduced Hangul keyboard portion, (A) the phonemes 'ㄱ' and 'ㄲ' are assigned to button '1', (B) the phonemes 'b' and 'ㅌ' are assigned to button '2', (C) the phonemes 'ㄷ', 'ㄸ', 'ㅓ' and 'ㅕ' are assigned to button '3', (D) Phonemes 'ㄹ' and 'ㅋ' are assigned to button '4', (E) Phonemes 'ㅁ', 'ㄷ', 'ㅡ' are assigned to button '5', (F) the phonemes 'ㅂ', 'ㅃ', 'ㅗ' and 'ㅛ' are assigned to button '6', (G) Phonemes 'ㅅ', 'ㅆ', 'ㅜ'. 'ㅠ' is assigned to button '7', (H) The phonemes 'ㅇ' and 'ㅊ' are assigned to button '8', (I) Phonemes 'ㅈ', 'ㅉ', 'ㅣ' are assigned to button '9', (T) Korean input device, characterized in that the phonemes 'ㅎ', 'ㅏ', 'ㅑ' are assigned to the button '0'. According to claim 1 or 2, wherein the reduced Hangul keyboard portion, (A) the phonemes 'ㄱ' and 'ㄲ' are assigned to button '1', (B) the phonemes 'b' and 'ㅌ' are assigned to button '2', (C) the phonemes 'ㄷ', 'ㄸ', 'ㅓ' and 'ㅕ' are assigned to button '3', (D) Phonemes 'ㄹ' and 'ㅋ' are assigned to button '4', (E) Phonemes 'ㅁ', 'ㄷ', 'ㅡ' are assigned to button '5', (F) the phonemes 'ㅂ', 'ㅃ', 'ㅏ' and 'ㅑ' are assigned to button '6', (G) Phonemes 'ㅅ', 'ㅆ', 'ㅗ'. 'ㅛ' is assigned to button '7', (H) The phonemes 'ㅇ' and 'ㅊ' are assigned to button '8', (I) Phonemes 'ㅈ', 'ㅉ', 'ㅣ' are assigned to button '9', (T) The Hangul input device, characterized in that the phonemes 'ㅎ', 'TT', 'ㅠ' are assigned to the button '0'. According to claim 1 or 2, wherein the reduced Hangul keyboard portion, (A) the phonemes 'ㄱ' and 'ㄲ' are assigned to button '1', (B) the phonemes 'b' and 'ㅌ' are assigned to button '2', (C) the phonemes 'ㄷ', 'ㄸ', 'ㅏ' and 'ㅑ' are assigned to button '3', (D) Phonemes 'ㄹ' and 'ㅋ' are assigned to button '4', (E) Phonemes 'ㅁ', 'ㄷ', 'ㅡ' are assigned to button '5', (F) the phonemes 'ㅂ', 'ㅃ', 'ㅗ' and 'ㅛ' are assigned to button '6', (G) Phonemes 'ㅅ', 'ㅆ', 'ㅜ'. 'ㅠ' is assigned to button '7', (H) The phonemes 'ㅇ' and 'ㅊ' are assigned to button '8', (I) Phonemes 'ㅈ', 'ㅉ', 'ㅣ' are assigned to button '9', (T) Korean input device, characterized in that the phonemes 'ㅎ', 'ㅓ', 'ㅕ' are assigned to the button '0'. According to claim 1 or 2, wherein the reduced Hangul keyboard portion, (A) Phonemes 'ㄱ', 'ㄲ' and 'ㅑ' are assigned to button '1', (B) Phonemes 'b', 'ㅌ' and 'ㅡ' are assigned to button '2', (C) the phonemes 'ㄷ', 'ㄸ' and 'ㅓ' are assigned to button '3', (D) The phonemes 'ㄹ', 'ㅋ' and 'ㅛ' are assigned to button '4', (E) Phonemes' ㅁ ',' '' and 'ㅗ' are assigned to button '5', (F) the phonemes 'ㅂ', 'ㅃ' and 'ㅏ' are assigned to button '6', (G) Phonemes 'ㅅ', 'ㅆ' and 'ㅕ' are assigned to button '7', (H) The phonemes 'ㅇ', 'ㅊ' and 'ㅠ' are assigned to button '8', (I) Phonemes 'ㅈ', 'ㅉ', 'ㅣ' are assigned to button '9', (T) Korean input device, characterized in that the phoneme 'ㅎ', 'TT' is assigned to the button '0'. According to claim 1 or 2, wherein the reduced Hangul keyboard portion, (A) Phonemes 'ㄱ', 'ㄲ' and 'ㅕ' are assigned to button '1', (B) the phonemes 'b', 'ㅌ' and 'ㅑ' are assigned to button '2', (C) the phonemes 'ㄷ', 'ㄸ' and 'ㅓ' are assigned to button '3', (D) The phonemes 'ㄹ', 'ㅋ' and 'ㅛ' are assigned to button '4', (E) Phonemes 'ㅁ', 'ㄷ', 'ㅡ' are assigned to button '5', (F) the phonemes 'ㅂ', 'ㅃ' and 'ㅏ' are assigned to button '6', (G) Phonemes 'ㅅ', 'ㅆ' and 'ㅗ' are assigned to button '7', (H) The phonemes 'ㅇ', 'ㅊ' and 'ㅠ' are assigned to button '8', (I) Phonemes 'ㅈ', 'ㅉ', 'ㅣ' are assigned to button '9', (T) Korean input device, characterized in that the phoneme 'ㅎ', 'TT' is assigned to the button '0'. According to claim 1 or 2, wherein the reduced Hangul keyboard portion, (A) Phonemes 'ㄱ', 'ㄲ' and 'ㅕ' are assigned to button '1', (B) the phonemes 'b', 'ㅌ' and 'ㅑ' are assigned to button '2', (C) the phonemes 'ㄷ', 'ㄸ' and 'ㅓ' are assigned to button '3', (D) The phonemes 'ㄹ', 'ㅋ' and 'ㅛ' are assigned to button '4', (E) Phonemes 'ㅁ', 't' and 'TT' are assigned to button '5', (F) the phonemes 'ㅂ', 'ㅃ' and 'ㅗ' are assigned to button '6', (G) Phonemes 'ㅅ', 'ㅆ' and 'ㅡ' are assigned to button '7', (H) The phonemes 'ㅇ', 'ㅊ' and 'ㅠ' are assigned to button '8', (I) phonemes 'ㅈ', 'ㅉ', 'ㅏ' are assigned to button '9', (T) Korean input device, characterized in that the phoneme 'ㅎ', 'ㅣ' is assigned to the button '0'. A phoneme button in which one or more Hangul phonemes are arranged in a consonant form in a reduced Hangul keyboard, a function button that is pressed to input dependent phonemes, and a determination that the candidate currently output to the liquid crystal display is determined as syllable strings for the undetermined button sequence. A button, a change button for outputting a next-order candidate to the liquid crystal display, a candidate selection button for moving to the candidate selection mode, and a delete button for canceling the most recently inputted phoneme button operation are assigned, and the operation mode is input for Hangul input. Switching to an input mode; Converting a code assigned to the button in response to receiving a button operation for selecting a button of the reduced Hangul keyboard; Receiving a phoneme button code from the reduced Hangul keyboard and adding a button code to an indeterminate button sequence, generating candidate syllable sequences corresponding to the indeterminate button sequence, and determining a ranking; Receiving a function button code from the reduced Korean keyboard unit and setting or releasing a function button flag in a toggle method; Determining a syllable sequence by receiving a definite button code from the reduced Hangul keyboard and initializing data used for the undetermined button sequence and analysis; Receiving a change button code from the reduced Hangul keyboard and setting a candidate selection mode flag if there are three or more candidates, and replacing two candidates displayed on the screen if there are two candidates; Receiving a delete button code from the reduced Hangul keyboard to delete the last element of the pending button sequence; Selecting and confirming a candidate to be outputted on the screen by receiving a button code from the reduced Hangul keyboard with a candidate selection mode flag set; Presenting a list of candidate syllable sequences on the screen with the candidate selection mode flag set; And And outputting the first-order candidate syllable sequence to a region corresponding to the indeterminate button sequence while the candidate selection mode flag is not set. 12. The method of claim 11, wherein the candidate syllable sequence is generated using a table-based parsing technique, and in order to reduce the amount of storage space, the syllable cell maintains a single syllable cell, an intermediate result cell and a final result cell in the parsing table. Hangul input method based on a non-deterministic phoneme input method characterized in that only after the intermediate result cell and the final result cell is maintained in the parsing table. The Hangul input method of claim 11, wherein the single syllable cell is configured by using a button string / syllable correspondence table while searching for a syllable corresponding to a button string. 12. The nondeterministic phoneme input method of claim 11, wherein the candidate syllable sequence is calculated based on a syllable context probability model, and the probability is used as a criterion for ranking candidate syllable sequences. Hangul input method. The Hangul input method of claim 11, wherein the syllable concatenation frequency considering the spacing from the corpus and the syllable concatenation frequency without the spacing are considered together in the process of extracting the syllable context probability. Way. 12. The method of claim 11, wherein in the process of extracting the syllable context probability, the syllable context probability data is used to measure the syllable context probability, the rank of the syllable context frequency, and the syllable context probability value. A Hangul input method based on a nondeterministic phoneme input method, which measures utilization and selects and uses only high probability probability data. 12. The method of claim 11, wherein a consensus concatenation distribution vector is obtained for each phoneme, a quantitative error occurrence estimator is calculated using a measure proportional to the magnitude of the individual vectors for the two vectors and inversely proportional to the cosine of the vector. And an alphabetical input method based on an indeterminate phoneme input method, characterized in that the optimum keyboard layout is automatically searched by selecting the arrangement of the minimum ambiguity resolution error occurrence estimates.