JP5050972B2 - Character input device - Google Patents

Description

  The present invention relates to a small character input device applied to a mobile terminal device or the like.

  Conventionally, in mobile terminal devices such as mobile phones and PDAs (Personal Digital Assistance), a so-called “kana turning (multi-tap)” method is often used to input characters. In the “kana turning” method, a plurality of characters are assigned to each of the numeric keys, and each time the numeric key is pressed, the assigned characters are sequentially switched so that a desired character can be designated. For example, kana characters (“ka”, “ki”, “ku”, “ke”, “ko”) and alphabets A to C are assigned to the numeric keypad “2”. Therefore, by pressing the numeric key a predetermined number of times, any of the above kana characters or alphabets can be displayed and the input can be confirmed.

However, the “kana turning” method has a fatal disadvantage that the number of times of pressing a key increases, and it takes much time to input information such as text. Therefore, in order to be able to input many characters, it is desirable to use a so-called full keyboard.
Japanese Patent Laid-Open No. 11-53093 JP 2007-280153 A

  However, portable terminal devices such as mobile phones and PDAs have a small surface area of the main body, and when a full keyboard is mounted, the key size for each character is remarkably reduced, making it difficult to input with a finger. A small full keyboard (but larger than the portable terminal device) premised on connection to the portable terminal device has also been put into practical use. However, when such a separate full keyboard is used, the portability of the device is significantly impaired.

  In Patent Document 1, a touch panel in which a large number of switches of minute resistive films are arranged, keyboard characters are virtually arranged on the touch panel, and a desired position on the touch panel is designated using an input pen or the like. An input device has been proposed. Further, Patent Document 2 proposes an input device that reduces input mistakes by displaying characters with high input frequency larger on a touch panel.

  Although what was described in patent document 1 made the input device small using a touch panel, the possibility that two or more characters may be pressed simultaneously still remains, and the problem of an erroneous input is not solved. Further, in Patent Document 2, a line (“A line”, “Ka line”) or the like is selected, and a character of a stage included in the line (for example, “K”, “K”, “Ki”, “ku”, “ke”, “ko”) are displayed according to the input frequency. However, the problem that several steps are required for the input is left as in the case of “kana turning”.

  It is an object of the present invention to provide an input device that reduces the number of key presses and reduces the possibility of erroneous input.

The character input device according to the present invention has a group of subkeys arranged in a matrix, and a character input unit in which a plurality of characters to be input are respectively written across a plurality of subkeys ,
When a character to be input by the character input means is operated by a user , a predicted value of which of the plurality of characters to be input is turned on for each of the turned on subkeys Subkey probability value calculating means for calculating a probability value for each character;
The probability value calculated for each character for each of the subkeys turned on by the subkey probability value calculating means is accumulated for each character, and this accumulated value is divided by the number of turned on subkeys, for each character. A physical probability value calculating means for calculating a physical probability value that is an average value of the probability values;
A character candidate list generating means for generating a character candidate list including the character and the physical probability value based on the physical probability value and storing the character candidate list in a storage means;
It is provided with.

  In a preferred embodiment, the subkey probability value calculating means stores, for each subkey, a character and a probability value that is a value predicted to be turned on to input the character in association with each other. With reference to the probability table, the probability value of each of the turned on subkeys is obtained.

  In another preferred embodiment, the subkey probability value calculating means obtains the probability value of each of the turned on subkeys based on the probability density function for each of the characters and the position of the turned on subkey. To do.

  In a preferred embodiment, the character candidate list means generates a character candidate list sorted in descending order of the physical probability value.

In a more preferred embodiment, a probability value for an output candidate character string that is a character string obtained by sequentially adding character candidates selected from the character candidate list, a probability value of a character candidate selected from a character candidate list, and Weighted probability value calculating means for calculating a weighted probability value obtained by multiplying a predetermined weight related to the character candidate selected from the character candidate list;
A new output candidate character string with the character candidate added to the end of the output candidate character string, and an output candidate character string list including the weighted probability value, and the output candidate character string stored in the storage unit List generating means.

  In a preferred embodiment, the weighting probability calculating means refers to the output candidate character string and the character candidate, and determines whether or not the output candidate character string and the character candidate can be combined. Have

  In another preferred embodiment, the weight is a value determined according to the frequency of pressing the character candidate.

The character input program according to the present invention includes a subkey group arranged in a matrix, and a computer having character input means in which a plurality of characters to be input are written across a plurality of subkeys. On the computer,
When a character to be input by the character input means is operated by a user , a predicted value of which of the plurality of characters to be input is turned on for each of the turned on subkeys A subkey probability value calculating step for calculating a probability value for each character;
The probability value calculated for each character for each of the subkeys turned on by the subkey probability value calculation step is accumulated for each character, and this accumulated value is divided by the number of turned on subkeys, for each character. A physical probability value calculating step for calculating a physical probability value that is an average value of the probability values;
A character candidate list generation step of generating a character candidate list including the character and the physical probability value based on the physical probability value and storing the character candidate list in a storage unit;
Is executed.

In a preferred embodiment, the computer includes:
A probability value for an output candidate character string that is a character string obtained by sequentially adding character candidates selected from the character candidate list, a probability value of a character candidate selected from the character candidate list, and a character candidate selected from the character candidate list A weighted probability value calculating step of calculating a weighted probability value obtained by multiplying a predetermined weight related to the character candidate;
A new output candidate character string with the character candidate added to the end of the output candidate character string, and an output candidate character string list including the weighted probability value, and the output candidate character string stored in the storage means And a list generation step.

  An object of the present invention is to provide an input device that reduces the number of key presses and reduces the possibility of erroneous input.

  Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a diagram illustrating an external appearance of a mobile terminal device according to a first embodiment of the present invention. In the present embodiment, the character input device is applied to a mobile terminal device (for example, a mobile phone) having a communication function. As shown in FIG. 1, the mobile terminal device 10 according to the present embodiment includes a lower main body 11 and an upper main body 12 that serves as a cover when closed. The lower body 11 and the upper body 12 are connected by a hinge 19. A character input unit 13 for character input is disposed at the center of the lower body 11. As will be described later, the character input unit 13 is composed of a plurality of subkeys arranged in a matrix. A display unit 14 including a liquid crystal display device is provided in the center of the upper body 12.

  The surface of the character input unit 13 is substantially flat, and each subkey has a switch element. A plurality of subkeys (3 × 3 subkeys in the present embodiment) indicate one character, and the plurality of subkeys basically correspond to the displayed characters. However, as will be described later, the correspondence between a plurality of subkeys and characters is not fixed in a one-to-one relationship. The possibility that a subkey other than the subkey was pressed is considered. Since the portable terminal device 10 according to the present embodiment is small, when the character input unit 13 is pressed with a finger, a relatively large number of subkeys are turned on as indicated by reference numeral 100, for example. Therefore, in the present embodiment, which character is highly likely to be designated is calculated based on the turned-on subkey and presented on the screen of the display unit 14 (see reference numeral 102). In addition, a list of character candidates that may be specified (character candidate list) can be generated, and the character candidate list can be displayed on the screen of the display unit 14.

  In addition, a plurality of switches 15, 16, 17, and 18 are disposed on the lower main body 11 of the mobile terminal device 10 according to the present embodiment, for example. These switches function as enter keys (CR keys) and function keys. Although not shown in FIG. 1, it goes without saying that a switch can also be arranged on the upper body 12.

  FIG. 2 is a block diagram showing a hardware configuration of the mobile terminal device according to the present embodiment. As illustrated in FIG. 2, the mobile terminal device 10 includes a CPU 21, a RAM 22, a ROM 23, a character input unit 13, a switch unit 24, a display unit 14, and a communication unit 25. The CPU 21 executes various processes such as overall control of the mobile terminal device 10, character input recognition based on the subkey turned on by the character input unit 13, and switch input recognition based on an operation of a switch included in the switch unit 24. The ROM 23 stores programs such as character input recognition processing and switch input recognition processing to be executed by the CPU 21. The ROM 23 stores a physical probability table and a language probability table that are referred to in the character input recognition process. The RAM 22 temporarily stores a program read from the ROM 23 and stores data generated in the process.

  The switch unit 24 includes switches 15 to 18 shown in FIG. The communication unit 25 includes, for example, a transmission / reception circuit that can be connected to a mobile phone network, and a communication circuit that can be connected to a wireless LAN base station.

  FIG. 3 is a diagram showing the character input unit according to this embodiment in more detail. As shown in FIG. 3, in the present embodiment, rectangular subkeys (see, for example, reference numerals 301 and 302) are arranged in a matrix. As described above, each of the sub keys has a switch element. In addition, a mark indicating a certain character (“Q” in the area defined by the reference numerals 311 and 312) is attached to the area of the 3 × 3 matrix (refer to the reference numerals 311 and 312). The subkeys constituting the 3 × 3 matrix and the characters attached to the area are basically associated with each other. As can be understood from the characters attached to the 3 × 3 matrix area, in the present embodiment, the character input unit 13 is the same as the alphabet arrangement in a general full keyboard.

  FIG. 4 is a flowchart showing an example of processing executed in the mobile terminal device according to the present embodiment. As shown in FIG. 4, the CPU 21 of the portable terminal device executes predetermined initialization processing such as initialization of the work area of the RAM 22 and initialization of the screen of the display unit 14 (step 401).

  The CPU 21 executes a character input recognition process for recognizing the input character and character string based on the turned-on subkey of the character input unit 13 (step 402). This character input recognition process will be described in detail later. Further, the CPU 21 executes a switch input recognition process in accordance with the switches other than the character input unit 13 (for example, the switches 15 to 18 shown in FIG. 1 are turned on (step 403). Next, the CPU 21 displays the character or character string recognized in the character input recognition process or a list thereof on the screen of the display unit 14 (step 404) In step 404, the switch input recognition process is performed. The processing result is also reflected, and a predetermined character, character string, or the like is displayed on the screen of the display unit 14. Thereafter, the CPU 21 performs other necessary processing, for example, data communication with the outside via the communication unit 25, or the like. Is executed (step 405).

  FIG. 5 is a flowchart showing an example of character input recognition processing according to the present embodiment. As shown in FIG. 5, the CPU 21 scans the subkeys of the character input unit 13 to determine whether all the subkeys are in the OFF state (step 501). If it is determined Yes in step 501, the CPU 21 sets a flag OF indicating that all subkeys are off to “1” (step 502). The value of this flag is stored in a predetermined area of the RAM 22. On the other hand, if it is determined No in step 501, that is, if any of the subkeys is on, the CPU 21 determines whether or not the flag OF is “1” (step 503). If it is determined No in step 503, the process ends. On the other hand, if it is determined as Yes at step 503, the processing after step 504 is further executed. With this flag OF, the processing after step 504 is executed when the sub key is first turned on.

  If YES is determined in step 503, the CPU 21 specifies the turned-on subkey, stores information for specifying the subkey (for example, the position of the subkey) in a predetermined area of the RAM 22 (step 504), and The flag OF is reset to “0” (step 505). The position of the sub key can be represented by coordinates, for example. In the character input unit shown in FIG. 3, the coordinates of the sub key 301 are (1, 1) and the coordinates of the sub key 302 are (2, 1) with the lower left as the origin. The coordinates of the lower right subkey 303 are (30, 1), and the coordinates of the upper left subkey 304 are (1, 9). Therefore, the CPU 21 may store the coordinates in the RAM 22.

  Next, the CPU 21 calculates a physical probability with reference to the turned-on subkey (step 506). In this embodiment, the probability value of physical probability (physical probability value) is a value that predicts for each character which character the user has attempted to input when a certain series of subkeys is pressed. . FIG. 6 is a flowchart illustrating an example of a physical probability calculation process according to the present embodiment. As shown in FIG. 6, the CPU 21 selects a subkey to be processed among the turned-on subkeys (step 601). FIG. 7A is a diagram illustrating an example of a sub key turned on in the character input unit 13. In FIG. 7A, the hatched subkey indicated by reference numeral 701 is turned on. There are 22 subkeys in the area indicated by reference numeral 701. In step 504 of FIG. 5, the 22 subkeys are specified, and these subkeys are sequentially selected in step 601, and the process proceeds.

  The CPU 21 acquires a probability value corresponding to the sub key from the physical probability table stored in the ROM 23 (step 602). Hereinafter, a physical probability table and a probability value that is a stored value of the table will be described. In the present embodiment, for each subkey, when the subkey is turned on, a probability value predicting which character is turned on to input it is combined with the character and stored in the physical probability table. Has been.

  FIG. 8A shows that the subkey at position (8, 5) is turned on, and FIG. 8B shows the physical probability table corresponding to the subkey at position (8, 5). It is a figure which shows the combination of the inside character and a probability value. The subkey at position (8, 5) is located approximately at the center of the display of the letter “D”. Therefore, in the physical probability table, the probability value of the character “D” is 36% in the combination of the character and the probability value associated with this subkey. In addition, the probability values of the characters adjacent thereto, for example, “S”, “E”, “F”, and “R” are 13%, 12%, 12%, and 10%, respectively.

  9A is a diagram showing that the subkey at the position (9, 4) is turned on, and FIG. 9B is a physical probability table corresponding to the subkey at the position (9, 4). It is a figure which shows the combination of the inside character and a probability value. The subkey at position (9, 4) is located slightly to the lower right from the center of the display of the letter “D”. Accordingly, in the physical probability table, in the combination of the character and probability value associated with this subkey, the probability value of the character “D” is 29%, and the character “D” for the subkey at position (8, 5). ”Is lower than the probability value. On the other hand, the probability value of the letter “F” adjacent to the right side is 24%, and the probability values of the letters “C” and “V” adjacent to the lower side are relatively high, 17% and 16%, respectively.

  Thus, in the physical probability table according to the present embodiment, combinations of characters and probability values are stored for the subkeys at each position. For a large number of subjects, when sub-keys are turned on when a predetermined character is to be specified, it accumulates which subkey was turned on, and a statistical numerical value is obtained from the accumulated information, and a probability value based on the statistical numerical value in advance. Is calculated and stored in the physical probability table.

  Next, the CPU 21 accumulates the probability value for each character (step 603). For example, in the example of FIGS. 8 and 9, the probability values 36% and 29% are added for the character “D”, and the probability values 13% and 5% are added for the character “S”. The CPU 21 determines whether or not acquisition of probability values (step 602) and accumulation of probability values for each character (step 603) have been completed for all the subkeys that have been turned on (step 604). If NO is determined in step 604, the process returns to step 601, and a new subkey to be processed is selected.

  On the other hand, if it is determined Yes in step 604, the CPU 21 divides the cumulative value of the probability value for each character by the total number of turned on subkeys to obtain the average value of the probability value for each character. A probability value (physical probability value) of the physical probability is calculated (step 605). The CPU 21 sorts the physical probability values as a character candidate list in descending order of the physical probability values and stores them in a predetermined area of the RAM 22 (step 606).

  FIG. 7B is a diagram showing an example of physical probability values when the subkey indicated by reference numeral 701 in FIG. 7A is in the on state. In the example shown in FIGS. 7A and 7B, the probability value for each character is accumulated for each of the 22 subkeys in the on state, and then the accumulated value is represented by the total number of subkeys “22”. The average value of the probability values is obtained for each character. The higher physical probability value is used to calculate the next language probability as an input character candidate.

  When the physical probability is calculated, the CPU 21 calculates a language probability (step 507). FIG. 10 is a flowchart showing an example of language probability calculation processing according to the present embodiment. The probability value of the language probability is a value obtained by predicting the probability of an output candidate character string that is a character string candidate input by the user when characters are sequentially input by the user.

  In FIG. 10, weighting is performed on a new output candidate character string obtained by adding the character candidate obtained in step 506 to the output candidate character string that is a character string candidate input by the user and adding the character candidate. A probability value is calculated. First, the CPU 21 reads a predetermined output candidate character string and a weighted probability value associated with the output candidate character string from the output candidate character string list stored in a predetermined area of the RAM 22 (step 1001). The output candidate character string list stores output candidate character strings and related weighted probability values in descending order of weighted probability values. Therefore, the CPU 21 sequentially reads out the output candidate character strings and the related weighted probability values from the output candidate character string list in descending order of the weighted probability values. Next, the CPU 21 selects a predetermined character candidate from the character candidate list, and generates a character string in which the character candidate is added to the end of the output candidate character string (step 1002). In the character candidate list, as described above, character candidates are listed in descending order of physical probability values. Therefore, in step 1002, the CPU 21 may select character candidates sequentially in descending order of physical probability values.

  The CPU 21 determines whether or not the combination of the output candidate character string and the character candidate can actually exist as a character string (step 1003). In the present embodiment, as a character input mode, a Roman character mode for inputting Roman characters, an alphanumeric mode for directly inputting alphanumeric characters, and a character input unit 13 which are not shown in the figure, but a kana character is directly used instead of the alphabet. There is an input kana mode, and it operates under any character input mode. In each case, it is determined whether or not a combination can exist as a character string. As an example, the determination process in the Roman character mode will be described in more detail. FIG. 18 is a flowchart illustrating an example of combination determination processing in the Roman character mode.

  As shown in FIG. 18, the CPU 21 first determines whether or not the last character of the output candidate character string is a vowel (step 1801). If YES in step 1801, that is, if the last character is a vowel, the CPU 21 determines whether the character candidate to be added to the output candidate character string is a consonant (step 1802). If it is determined as Yes in step 1802, that is, if the character candidate is a consonant, it is determined as a possible combination (Yes in step 1003 in FIG. 10), and the process proceeds to step 1004. That is, when a consonant comes after a vowel, it is determined that there is a high possibility of making a meaning as a character string.

On the other hand, if it is determined No in step 1802, that is, if the end of the output candidate character string and two vowels are consecutive in the character candidate, the CPU 21 refers to a plurality of consecutive vowels. Thus, it is determined whether or not the continuous vowels are valid as notations (step 1803). More specifically, the following rules may be stored in the ROM 23 and the rules stored in the ROM 23 may be referred to. For example, the rules include
(1) Two consecutive vowels are valid (2) Three consecutive vowels are invalid only when all three are the same, otherwise valid (3) In the case of four consecutive vowels, the back side Logic can be included that is invalid only when all three are the same, and valid otherwise.

  If it is determined as Yes in step 1803, it is determined as a possible combination (Yes in step 1003 in FIG. 10), and the process proceeds to step 1004. On the other hand, if it is determined No in step 1803, the process proceeds to step 1006.

  If it is determined No in step 1801, that is, if the last character of the output candidate character string is a consonant, the CPU 21 determines whether the character candidate to be added to the output candidate character string is a vowel. (Step 1804). If it is determined Yes in step 1804, that is, if the character candidate is a vowel, it is a possible combination (Yes in step 1003 in FIG. 10), and the process proceeds to step 1004. That is, when a vowel comes after a consonant, it is determined that there is a high possibility of making a meaning as a character string.

On the other hand, when it is determined No in step 1804, that is, when two consonants are continuous at the end of the output candidate character string and the character candidates, the CPU 21 refers to a plurality of consecutive consonants, It is determined whether or not the consecutive consonants are valid as notations (step 1805). More specifically, the following rules may be stored in the ROM 23 and the rules stored in the ROM 23 may be referred to. For example, the rules include
(1) If two consecutive consonants correspond to any of the Hebon notations, the two consecutive consonants are valid,
(2) When two consecutive consonants correspond to a prompt sound, the two consecutive consonants are valid,
(3) If three consecutive consonants are combined with a prompting sound and a Hebon expression, the logic that the three consecutive consonants are valid may be included.

  If it is determined as Yes in step 1805, the combinations are possible (Yes in step 1003 in FIG. 10), and the process proceeds to step 1004. On the other hand, when it is determined No in step 1805, the CPU 21 further determines whether or not the continuous consonant corresponds to the abbreviation (step 1806). In step 1806, for example, an abbreviation dictionary may be stored in the ROM 23, and it may be determined whether or not consecutive consonants and abbreviations contained in the abbreviation dictionary match by a so-called forward matching search. If it is determined as Yes in step 1806, it is a possible combination (Yes in step 1003 in FIG. 10), and the process proceeds to step 1004. On the other hand, if it is determined No in step 1806, the process proceeds to step 1006.

  In the case of kana characters, for example, the following combination determination process may be executed. A word dictionary is prepared in the ROM 23, and it is only necessary to determine whether or not the character string matches the word by so-called forward matching search. Also in the case of alphanumeric input, a forward matching search using a word dictionary may be performed, or the combination determination process may be omitted.

  When the combination determination process ends, the CPU 21 calculates a weighted probability value of a new output candidate character string in which character candidates are added to the end of the output candidate character string (step 1004). The calculation of the weighted probability will be described in detail later. The CPU 21 stores the new output candidate character string and the associated weighted probability value as elements of the new output candidate character string list in a predetermined area of the RAM 22 (step 1005).

  Next, the CPU 21 determines whether or not the processes in steps 1003 to 1005 have been executed for all the character candidates included in the character candidate list (step 1006). If it is determined No in step 1006, the processing from step 1002 onward is executed for the next character candidate in the character candidate list. When it is determined YES in step 1006, the CPU 21 determines whether or not the processing in steps 1001 to 1006 has been executed for all output candidate character strings in the output candidate character string list (step 1007). ). If it is determined No in step 1007, the processing from step 1001 onward is executed for the next output candidate character string in the output candidate character string list. If it is determined YES in step 1007, the language probability calculation process ends.

Next, the calculation of the weighted probability value (step 1004) will be described in more detail with reference to FIGS. In the present embodiment, the subkeys in the character input unit 13 are pressed in the order of FIG. 11A, FIG. 12A, FIG. 13A, FIG. 14A, and FIG. I think.
Initially, as shown in FIG. 11A, the subkey in a certain area 1101 is pressed in the character input unit 13. At this time, the character candidate list includes values of K, L, I, J, and U in descending order of physical probability values, as in the left side of FIG. The physical probability values associated with each are P _K , P _L , P _I , P _J , and P _U. In the first language probability calculation process, the character candidate list is directly used as the output candidate character string list.

Next, as shown in FIG. 12A, the subkey in the area 1201 is pressed in the character input unit 13. In this case, as shown in FIG. 12B, the character candidate list includes values of Z, A, Q, and X in descending order of physical probability values, and the corresponding physical probability values are P _{Z, P} a, _P Q, the _{P X.} In the language probability calculation processing, it is determined whether or not possible combinations are possible between the output candidate character strings K to U and the character candidates Z, A, Q, and X (see step 1003). The weighted probability values are calculated (see step 1004). As shown in FIG. 12C, the output candidate character strings are added to the new output candidate character string list in descending order of the probability values (weighted probability values). Be placed.

In the present embodiment, the weighted probability value P of the new output candidate character string is
P = P _i-1 × P _i × W _k
Is calculated by Here, P _i-1 is a weighted probability value for the original output candidate character string, P _i is a probability value of a character candidate added to the output candidate character string, and W _k is the original output candidate. The weight is determined by the relationship between the character string and the character candidate. In the present embodiment, for example, n stages of weights W _k (k = 1, 2,..., N) are provided for each character based on the frequency of pressing the character, and the character is associated with the character. The weights thus obtained are stored in the ROM 23 as a weight table.

For example, in FIG. 12C, the weighted probability values P _KA , P _IA , P _IZ , P _IQ , P _JA and P _UZ of the output candidate character strings KA, IA, IZ, IQ, JA and _UZ are as follows: Is calculated.
P _KA = P _K × P _A × W _A (W _A is a weight associated with the letter A)
_{P IA = P I × P A} × W A
P _IZ = P _I × P _Z × W _Z (W _Z is the weight associated with the letter Z)
P _IQ = P _I × P _Q × W _Q (W _Q is a weight associated with the letter Q)
_{P JA = P J × P A} × W A
P _UZ = P _U × P _Z × W _Z

In the example of FIG. 13A, in the character input unit 13, the subkey in the area 1301 is pressed, a character candidate is added to the output candidate character string of 2 characters, and a new output candidate character string of 3 characters is created. It is done. FIG. 13B shows the generated character candidate list. FIG. 13C shows the generated new output candidate character string and the weighted probability value. In FIG. 13 (c), the output candidate character strings KAE, KAW, KAR, KAD, IZE and KAS weighted probability value _P KAE _{of, P KAW, P KAR, P} KAD, P IZE and _{P KAS} is calculated as follows Is done.
_{P KAE = P KA × P E} × W E (W E is the weight associated with the character E)
P _KAW = P _KA × P _W × W _W (W _W is a weight associated with the letter W)
_{P KAR = P KA × P R} × W R (W R is the weight associated with the character R)
P _KAD = P _KA × P _D × W _D (W _D is the weight associated with the letter D)
P _IZE = P _IZ × P _E × W _E
P _KAS = P _{K A} × P _S × W _S (W _S is the weight associated with the letter S)

In the example of FIG. 14A, in the character input unit 13, the subkey in the area 1401 is pressed, character candidates are added to the output candidate character string of 3 characters, and a new output candidate character string of 4 characters is created. It is done. FIG. 14B shows a generated character candidate list. FIG. 14C shows the generated new output candidate character string and the weighted probability value. In FIG. 14C, the weighted probability values P _KAER , P _KAEE , P _KAEF , P _KAED , P _IZER and P _KAWE of the output candidate character strings KAER, KAEE, KAEF, KAED, IZER and KAWE are calculated as follows. Is done.
_{P KAER = P KAE × P R} × W R
P _KAEE = P _KAE × P _E × W _{E
P KAEF = P KAE × P F} × W F
P _KAED = P _KAE × P _D × W _{D
P IZER = P IZE × P R} × W R
P _KAWE = P _KAW × P _E × W _E

In the example of FIG. 15A, in the character input unit 13, the subkey in the area 1501 is pressed, character candidates are added to the 4-character output candidate character string, and a new 5-character output candidate character string is created. It is done. FIG. 15B shows the generated character candidate list. FIG. 15C shows the generated new output candidate character string and the weighted probability value. In FIG. 15C, the weighted probability values P _KAERU , P _KAERI , P _KAEUU , P _KAWEU and P _KAEDU of the output candidate character strings KAERU, KAERI, KAEEU, KAWEU and KAEDU are calculated as follows.
P _KAERU = P _KAER × P _U × W _U
P _KAERI = P _KAER × P _I × W _I
P _KAEEEU = P _KAEE × P _U × W _U
P _KAWE = P _KAWE × P _U × W _U
P _KAEDU = P _KAED × P _U × W _U

  As described above, as a character is input to the character input unit 13, a character candidate list is generated, and a character candidate in the character candidate list is added to the output candidate character string of the output candidate character string list. A new output candidate character string is generated. Also, a weighted probability value is calculated for the new output candidate character string. A list that includes a weighted probability value associated with the final output candidate character string that is greater than a predetermined value, or a list that includes a predetermined number of output candidate character strings selected from the top of the weighted probability value ( In step 404) of FIG. 4, it can be displayed on the screen of the display unit 14. Alternatively, an output candidate character string having the highest weighted probability value may be displayed on the screen of the display unit 14.

  Next, the switch input recognition process (step 403 in FIG. 4) according to the present embodiment will be described in more detail. 16 and 17 are flowcharts illustrating an example of the switch input recognition process according to the present embodiment. As shown in FIG. 16, the CPU 21 checks the operation status of the switches (for example, the switches 15 to 18) constituting the switch unit 24, and determines whether or not there is a function change (step 1601). If it is determined Yes in step 1601, the CPU 21 determines whether or not a character input is necessary when changing the function (step 1602). If YES in step 1602, that is, if character input is required, the CPU 21 sets the function flag KF to “1” (step 1603), whereas if NO is determined in step 1602, the CPU 21 Resets the function flag KF to “0” (step 1604).

  Next, the CPU 21 determines whether or not the function flag KF is “1” (step 1605). If NO in step 1605, the process proceeds to step 1709. If it is determined Yes in step 1605, the CPU 21 refers to the operation of the switches constituting the switch unit 24 and determines whether or not the character input mode has been switched (step 1606). If it is determined No in step 1606, the process proceeds to step 1701 in FIG.

  On the other hand, if it is determined Yes in step 1606, the CPU 21 determines what the character input mode is (step 1607), and has entered any one of the alphanumeric mode, the kana mode, and the Roman character mode. Accordingly, the character input mode flag NF is set to “0”, “1”, and “2”, respectively (steps 1608 to 1610). If there is no change in the character input mode (No in step 1606), the same value is maintained in the character input mode flag NF. Subsequently, the CPU 21 determines whether or not the character input mode flag NF is “0”, that is, whether or not the character input mode is an alphanumeric mode (step 1701). If it is determined YES in step 1701, the process proceeds to step 1707. This is because Kanji conversion by the conversion key is not performed in the alphanumeric mode.

  If it is determined No in step 1701, that is, if the character input mode is the Roman character mode or the kana mode, the CPU 21 determines whether or not the switch assigned as the conversion key has been pressed in the switch unit 24. (Step 1702). If it is determined No in step 1702, the process proceeds to step 1705. If YES is determined in step 1702, the CPU 21 further determines whether there is an unconfirmed character string (step 1703). If it is determined No in step 1703, the process proceeds to step 1707. If it is determined Yes in step 1703, the CPU 21 executes kanji conversion processing (step 1704). The kanji conversion process is the same as that executed by a conventional word processor or the like, and the CPU 21 matches the reading based on the character string (in this embodiment, the output candidate character string selected by the user). Generate a list of kanji like The generated list is displayed on the screen of the display unit 14 in the display process (step 404 in FIG. 4).

  Next, the CPU 21 determines whether or not a switch assigned as a confirmation key has been pressed in the switch unit 24 (step 1705). When it is determined Yes in step 1705, the CPU 21 stores the kanji string converted from the character string at the time when the confirmation key is pressed in a predetermined area of the RAM 22 as a confirmed word ( Step 1706).

  Further, the CPU 21 determines whether or not a predetermined switch is operated in the switch unit 24 to instruct switching of the output candidate character string (step 1707). When it is determined Yes in step 1707, the CPU 21 refers to the output candidate character string list and selects an output candidate character string having the next highest weighted probability value (step 1708). The newly selected output candidate character string is displayed on the screen of the display unit 14 in the display process (step 404 in FIG. 4). For example, when the confirmation key is not pressed, the currently selected output candidate character string can be converted by the conversion key. Accordingly, when switching of the output candidate character string is instructed in a state where it can be converted, it becomes possible to select another output candidate character string in step 1708 as described above. Even in the alphanumeric mode, the candidate switching determination (step 1707) and the candidate switching process (step 1708) are executed.

  Thereafter, the CPU 21 executes processing associated with another switch input (step 1709). This processing corresponds to, for example, switch input for calling a function other than character input, for example, call of an image or selection of a destination at the time of communication.

  In FIG. 4, the character candidate list and the output candidate character string list generated in the character input recognition process (step 402) are displayed on the screen of the display unit 14 in the display process (step 404). In addition, the kanji converted in the switch input recognition process (step 403) is also displayed on the screen of the display unit 14 in the display process.

  According to the present embodiment, when any of the sub keys is turned on, the CPU 21 calculates a probability value that is a predicted value of which character is turned on for each of the turned on sub keys. Calculated for each character, accumulates the probability value calculated for each character, and based on the accumulated value, a physical value that is a prediction of whether the character was turned on to input that character for the entire turned on subkey Probability value is calculated. Thereby, when a series of subkeys are pressed, it is possible to predict which character the user has turned on in order to input the series of subkeys, and to specify the character candidates. Thereby, even if the character input means is small and takes the form of a full keyboard, the possibility of inputting characters intended by the user can be remarkably increased. Further, there is no need for complicated input such as turning a kana.

  Further, the CPU 21 generates a character candidate list including characters and physical probability values based on the physical probability values, and stores them in the RAM 22. By displaying this character candidate list on the screen of the display means and presenting it to the user, it is possible to allow the user to specify the character that the user intends to input.

  In the first embodiment, for each subkey, refer to a physical probability table in which a character and a probability value that is a value predicted to be turned on to input the character are associated and stored. , Get the probability value of each of the turned subkeys. Thereby, it is possible to eliminate the need for a complicated calculation when calculating the probability value.

  In the first embodiment, the character candidate list is sorted in descending order of physical probability values. As a result, when the character candidate list is displayed on the screen of the display device, the characters that are predicted to be input by the user can be shown higher.

  In the first embodiment, the CPU 21 also sets the probability value for the output candidate character string, which is a character string obtained by sequentially adding the character candidates selected from the character candidate list, and the probability of the character candidate selected from the character candidate list. A new output candidate character string obtained by calculating a weighted probability value obtained by multiplying the value and a predetermined weight related to the character candidate selected from the character candidate list, and adding the character candidate to the end of the output candidate character string, An output candidate character string list including weighted probability values is generated and stored in the RAM 22. Thereby, it is possible to remarkably increase the possibility of inputting a character string according to the user's intention with respect to the character strings sequentially input. Further, by displaying the output candidate character string list on the screen of the display means and presenting it to the user, the user can designate the character string that the user intends to input.

  Further, the CPU 21 refers to the output candidate character string and the character candidate and determines whether or not the output candidate character string and the character candidate can be combined. As a result, combinations in which character strings cannot be chained can be eliminated.

  In the first embodiment, the weight is a value determined according to the frequency of pressing the character candidate. As a result, a weighted probability value with a greater weight can be obtained for a character that is frequently pressed by the user, and a more appropriate output candidate character string can be obtained.

  Next, a second embodiment of the present invention will be described. In the first embodiment, a physical probability table storing combinations of characters and probability values for each subkey is stored in the ROM 23, and the CPU 21 acquires physical probability values by referring to the physical probability table. On the other hand, in the second embodiment, a physical probability value is acquired using a Gaussian distribution. When using the Gaussian distribution, there are a case where a table is used and a case where a function is calculated. In the second embodiment, a physical probability value is obtained by calculating a Gaussian distribution.

  Hereinafter, the calculation principle of the physical probability value according to the second embodiment will be described. The Gaussian distribution is a distribution having a probability density function expressed by Formula 1.

μ is an average, and σ is a standard deviation. This probability density function becomes a curve as shown in FIG. In FIG. 19, the x axis (horizontal axis) is considered as the horizontal axis where the characters of the character input unit 13 are arranged, and the position (coordinates) of μ is associated with the position of the character indicated in the character input unit 13. It becomes possible. In the example of FIG. 19, among the three horizontal axes of the character input unit 13, the uppermost horizontal axis (that is, the horizontal axis where the characters Q, W, E, R,... Are arranged in order from the left). And the x-axis are associated with each other, and the character R is associated with the position of μ. That is, this can be considered as a Gaussian distribution for the letter R. Therefore, by assigning subkeys on the x-axis in the same manner as the above characters, a probability value for a character (character R in FIG. 19) can be acquired for each subkey. For example, the probability value for the character R of the subkey at the center (position x ₀ ) of the character R can be obtained by f (x ₀ ). Similarly, the probability values for the character R of the subkeys on both sides (positions x ₁ , x ₋₁ ) in the horizontal axis direction are f (x ₁ ) = f (x ₋₁ ), respectively. Probability value of the position _{x 2} adjacent lateral position _{x 1} is f _{(x 2),} which is smaller than f _{(x 1).} In this way, based on the Gaussian distribution, it is possible to obtain a probability value that gradually decreases as the distance from the average μ increases.

  Actually, the subkeys are two-dimensionally arranged in a matrix. Accordingly, as shown in FIG. 19, the Gaussian distribution includes not only the x axis but also the x axis and the y axis, and the probability density function f (x, y) is determined by x and y. FIG. 20 is a graph showing an example of a two-dimensional probability density function. Similarly to the example of FIG. 19, the character R is associated with the position of the average μ, and the subkey is assigned to a predetermined position on the x-axis and the y-axis, so that the probability value f (x , Y).

  FIG. 21 is a flowchart illustrating an example of a physical probability calculation process according to the second embodiment. In FIG. 21, steps 2102, 2106 to 2109 correspond to steps 601 and 603 to 606 in FIG. As can be understood from FIGS. 6 and 21, in the second embodiment, instead of acquiring the probability value from the physical probability table (step 602 in FIG. 6), the value of the probability density function is calculated ( Steps 2103 to 2105 in FIG.

  As shown in FIG. 21, the CPU 21 determines the standard deviation σ of the probability density function (step 2101). This may be determined according to user input, or a predetermined value may be used. Next, the CPU 21 selects a subkey to be processed among the turned on subkeys (step 2102). The CPU 21 specifies a character for which a probability value is to be calculated, and calculates μ for the character (step 2103). μ can be calculated based on the coordinates of the position (the center) where the character is indicated in the character input unit 13.

  Further, the CPU 21 calculates a function value f (x, y) of the probability density function of the subkey coordinates (x, y) (step 2104). The calculated function value f (x, y) is stored in the RAM 22. If processing has not been completed for all characters for which probability values are to be obtained (No in step 2105), the process returns to step 2103. Since the probability density function has a value “0” when x and y are “∞” or “−∞”, theoretically, probability values are calculated for all characters shown in the character input unit 13. There is a need. However, in practice, it is only necessary to process only characters included in a certain range from the subkeys to be processed.

  The CPU 21 accumulates the probability value for each character (step 2106). Further, the CPU 21 determines whether or not the acquisition of probability values (steps 2103 to 2105) and the accumulation of probability values for each character (step 2106) have been completed for all the subkeys that have been turned on (step 2107). . If NO is determined in step 2107, the process returns to step 2102 and a new subkey to be processed is selected.

  If it is determined Yes in step 2107, the CPU 21 divides the accumulated value of the probability value for each character by the total number of turned on subkeys to obtain the physical probability value that is the average value of the probability values for each character. Is calculated (step 2108). The CPU 21 sorts the physical probability values for each character as a character candidate list in descending order of physical probability values and stores them in a predetermined area of the RAM 22 (step 2109). The character candidate list including the characters and the calculated physical probability value is used for language probability calculation processing, as in the first embodiment.

  According to the second embodiment, the probability value of each of the turned-on subkeys is acquired based on the probability density function for each character and the position of the turned-on subkey. Therefore, although the calculation of the probability density function is necessary, the amount of data to be retained can be significantly reduced.

  The present invention is not limited to the above embodiments, and various modifications can be made within the scope of the invention described in the claims, and these are also included in the scope of the present invention. Needless to say.

  For example, in the above embodiment, the sub key is a switch that outputs ON / OFF, but it may be configured to output a plurality of values according to the pressed state (for example, pressed). That is, the pressing value of the sub key can take an intermediate value between a value indicating the state where the key is not pressed (for example, “0”) and a value indicating the state where the key is pressed most strongly (maximum value). In this case, even if the probability value is corrected by multiplying the probability value for each character corresponding to the subkey by a correction value (for example, the pressing value / maximum value) according to the pressing value of the subkey in step 602. good. As a result, the probability value of the sub key pressed more strongly is reflected more strongly in the whole.

  In the first embodiment, not only the character but also other characters in the center of the 3 × 3 matrix that is basically associated with a character in the physical probability table. Is stored in consideration of the possibility of trying to input (for example, see FIGS. 8A and 8B). However, the probability value is not limited to the first embodiment. For example, for the subkey at the center of the 3 × 3 matrix, the value of the physical probability table may be stored so that the probability value of the basically associated character is 100%. For example, in the examples of FIGS. 8A and 8B, the probability value associated with the letter D is “100%”. With such a configuration, if only the center subkey is pressed with a thin member such as a touch pen, the physical probability value of the input character can be set to 100%, and the character can be determined more easily. be able to. Of course, both the physical probability table as in the first embodiment and the other physical probability table described above may be stored in the ROM 23 and switched by operating a switch.

  In the above-described embodiment, the weighting probability is a value based on the frequency of pressing the character, but is not limited to this. For example, it is good also as a chain probability value which shows the high possibility of the chain | linkage between the last character of an output candidate character string, and the character candidate added. In this case, the combination determination process as shown in FIG. 18 can be omitted by setting the chain probability value to “0” for combinations that cannot exist.

  Furthermore, in the second embodiment, a probability density function of Gaussian distribution is tabulated for each character, a combination of coordinates and function values is stored in the ROM 23, and the function stored with reference to the table stored in the ROM 23 is used. It is good also as a structure which acquires a value.

FIG. 1 is a diagram illustrating an appearance of a mobile terminal device according to the present embodiment. FIG. 2 is a block diagram showing a hardware configuration of the mobile terminal device according to the present embodiment. FIG. 3 is a diagram showing the character input unit according to this embodiment in more detail. FIG. 4 is a flowchart showing an example of processing executed in the mobile terminal device according to the present embodiment. FIG. 5 is a flowchart showing an example of character input recognition processing according to the present embodiment. FIG. 6 is a flowchart illustrating an example of a physical probability calculation process according to the present embodiment. FIG. 7A is a diagram illustrating an example of a sub key turned on in the character input unit 13, and FIG. 7B is a diagram illustrating a calculated physical probability value. FIG. 8A shows that the subkey at position (8, 5) is turned on, and FIG. 8B shows the physical probability table corresponding to the subkey at position (8, 5). It is a figure which shows the combination of the inside character and a probability value. 9A is a diagram showing that the subkey at the position (9, 4) is turned on, and FIG. 9B is a physical probability table corresponding to the subkey at the position (9, 4). It is a figure which shows the combination of the inside character and a probability value. FIG. 10 is a flowchart showing an example of language probability calculation processing according to the present embodiment. FIG. 11A is a diagram showing a state where a sub key of a certain area is pressed, and FIG. 11B is a diagram showing an output candidate character string list at that time. These are the flowcharts which show the example of the automatic accompaniment performance process concerning this Embodiment. 12A shows a state where a subkey of a certain area is pressed, FIG. 12B shows a character candidate list at that time, and FIG. 12C shows a new output candidate character string. It is a figure which shows a list. 13A shows a state where a subkey of a certain area is pressed, FIG. 13B shows a character candidate list at that time, and FIG. 13C shows a new output candidate character string. It is a figure which shows a list. FIG. 14A shows a state where a subkey of a certain area is pressed, FIG. 14B shows a character candidate list at that time, and FIG. 14C shows a new output candidate character string. It is a figure which shows a list. 15A is a diagram showing a state where a subkey of a certain area is pressed, FIG. 15B is a diagram showing a character candidate list at that time, and FIG. 15C is a new output candidate character string. It is a figure which shows a list. FIG. 16 is a flowchart illustrating an example of switch input recognition processing according to the present embodiment. FIG. 17 is a flowchart illustrating an example of switch input recognition processing according to the present embodiment. FIG. 18 is a flowchart showing an example of combination determination processing in the Roman character mode according to the present embodiment. FIG. 19 is a graph for explaining the use of the probability density function in the second embodiment. FIG. 20 is a graph for explaining the use of the probability density function in the second embodiment. FIG. 21 is a flowchart illustrating an example of a physical probability calculation process according to the second embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Portable terminal device 11 Lower side body 12 Upper side body 13 Character input part 14 Display part 15-18 Switch 21 CPU
22 RAM
23 ROM
24 switch unit 25 communication unit

Claims (9)

Character input means having a group of subkeys arranged in a matrix, wherein a plurality of characters to be input are written across a plurality of subkeys ,
When a character to be input by the character input means is operated by a user , a predicted value of which of the plurality of characters to be input is turned on for each of the turned on subkeys Subkey probability value calculating means for calculating a probability value for each character;
The probability value calculated for each character for each of the subkeys turned on by the subkey probability value calculating means is accumulated for each character, and this accumulated value is divided by the number of turned on subkeys, for each character. A physical probability value calculating means for calculating a physical probability value that is an average value of the probability values;
A character candidate list generating means for generating a character candidate list including the character and the physical probability value based on the physical probability value and storing the character candidate list in a storage means;
A character input device comprising:   With reference to the physical probability table in which the subkey probability value calculating means stores a character and a probability value that is a value predicted to be turned on to input the character for each subkey, The character input device according to claim 1, wherein a probability value of each of the turned on subkeys is acquired.   The subkey probability value calculating means acquires a probability value of each of the turned on subkeys based on a probability density function for each character and a position of the turned on subkey. The character input device according to 1.   The character input device according to any one of claims 1 to 3, wherein the character candidate list means generates a character candidate list sorted in descending order of the physical probability values. A probability value for an output candidate character string that is a character string obtained by sequentially adding character candidates selected from the character candidate list, a probability value of a character candidate selected from the character candidate list, and a character candidate selected from the character candidate list Weighted probability value calculating means for calculating a weighted probability value obtained by multiplying a predetermined weight related to the character candidate;
A new output candidate character string with the character candidate added to the end of the output candidate character string, and an output candidate character string list including the weighted probability value, and the output candidate character string stored in the storage means A list generation means;
The character input device according to any one of claims 1 to 4, further comprising:   The weighting probability calculation means includes means for referring to the output candidate character string and the character candidate to determine whether or not a combination of the output candidate character string and the character candidate is possible. Item 6. The character input device according to Item 5.   The character input device according to claim 5, wherein the weight is a value determined according to a frequency of pressing the character candidate. In a computer having subkey groups arranged in a matrix and having character input means in which a plurality of characters to be input are written across a plurality of subkeys, the computer includes:
When a character to be input by the character input means is operated by a user , a predicted value of which of the plurality of characters to be input is turned on for each of the turned on subkeys A subkey probability value calculating step for calculating a probability value for each character;
The probability value calculated for each character for each of the subkeys turned on by the subkey probability value calculation step is accumulated for each character, and this accumulated value is divided by the number of turned on subkeys, for each character. A physical probability value calculating step for calculating a physical probability value that is an average value of the probability values;
A character candidate list generation step of generating a character candidate list including the character and the physical probability value based on the physical probability value and storing the character candidate list in a storage unit;
A character input program characterized in that In the computer,
A probability value for an output candidate character string that is a character string obtained by sequentially adding character candidates selected from the character candidate list, a probability value of a character candidate selected from the character candidate list, and a character candidate selected from the character candidate list A weighted probability value calculating step of calculating a weighted probability value obtained by multiplying a predetermined weight related to the character candidate;
A new output candidate character string with the character candidate added to the end of the output candidate character string, and an output candidate character string list including the weighted probability value, and the output candidate character string stored in the storage unit A list generation step;
The character input program according to claim 8 , wherein: