JP2000276466A - Device and method for predicting character string and storage medium stored with character string predicting program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make a prediction result easy for a user to understand and to lighten the load of input operation by converting only a partial character string before a prediction start position from KANA (Japanese syllabary) to KANJI (Chinese character) and presenting it additionally before a prediction candidate. SOLUTION: A syllabary string to be predicted in information inputted from an input part 101 is sent to a prediction control part 104 through a control part 102. The prediction control part 104 generates a character string group for retrieval including the character at the tail of the syllabary string by a retrieval character string generation part 105 and a prediction dictionary retrieval part 106 retrieves index data that the syllbary string corresponds to as a perdition candidate by referring to a prediction dictionary 107. Then a KANA-KANJI conversion control part 114 performs a KANA-KANJI converting process on the partial character string except the prediction candidate in the syllabary string to be predicted using a KANA-KANJI conversion part 108 to obtain a KANA-KANJI mixed character string candidate and sends it as a conversion candidate back to the control part 102 together with the prediction candidate. The sent-back KANA-KANJI conversion candidate is outputted through an output part 103 and displayed on a display, etc.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is used in a document creation device such as a personal computer or a word processor, and is intended to be subsequently input based on a character string input by an input device such as a keyboard. The present invention relates to a character string prediction device, a character string prediction method, and a storage medium that stores a character string prediction program.

[0002]

2. Description of the Related Art In recent years, networks have been rapidly developed,
Communication using computers is becoming popular. Along with this, in addition to the traditional main uses of document creation, such as creating business documents using a desktop computer and writing a dissertation, more personal use, such as using e-mail using a portable information device and creating memos, has been made. Even in unusual situations, the chances of entering Japanese sentences with a computer are increasing, and even people who are not skilled in keyboard operation are required to have a technique for accurately inputting information with few key touches. On the other hand, various electronic devices such as computers have become smaller, and portable devices have been widely used. For mobile devices,
Since the size of the input operation unit is small, the burden of inputting a document is further increased, and it is desired to reduce the burden associated with user input.

One solution to such a request is to input the beginning of a phrase to be entered, search for the phrase as a key, and present the phrase by inputting it (Japanese Patent Laid-Open No. 60-105029). ) Or a word processor (Japanese Patent Laid-Open Publication No. 1-260568) for searching for and presenting a word connected to a word whose conversion has been confirmed, inputting the head of the reading to be input, and searching for and presenting the reading to be input using the key as a key. Kanji conversion system (Japanese Patent Laid-Open No. 8-1472)
No. 89) has been proposed.

[0004] In the above-mentioned Japanese Patent Application Laid-Open No. 60-105029, the frequency information of reading idioms is stored, and the ranking of candidates is performed based on the information. Also, Japanese Patent Application Laid-Open No. 1-26056
In Japanese Patent Publication No. 8, priorities are assigned to candidates based on past input histories. Further, in Japanese Patent Application Laid-Open No. 8-147289, frequency information of conversion candidate words is stored, and the priorities of the candidates are assigned based on the frequency information.

However, all of these methods are based on the premise that the beginning of the character string to be input is input, and use this as a key to perform a forward match search of the dictionary to perform prediction. There is no mechanism for automatically predicting the input that follows at an arbitrary position inside. Therefore, when using during the input in the continuous phrase conversion method that has become popular in recent years, press the conversion key once in the middle of the conversion, interrupt the input operation, and then enter the beginning of the character string you want to input again, Further, key operations such as instructing a prediction search at a position where prediction is desired are required, and the operation is complicated.

On the other hand, in order to perform prediction from a position other than the head of the input character string, it is necessary to develop a process relating to a partial character string in front of the prediction start position. In order to solve this problem, a Japanese input prediction method has been proposed in which an input character string and a predicted and complemented character string are converted to kana-kanji characters, and the obtained conversion result is added before a prediction candidate and output. I have.

However, in this method, since the entire character string from the beginning of the input including the predicted and completed reading character string is converted to Kana-Kanji characters, the prediction of the Kanji notation character string is performed instead of the prediction of the reading notation character string. In this case, it is necessary to return to reading again up to the portion already written in kanji and perform kana-kanji conversion, which is wasteful. In addition, it is not efficient because it is necessary to ensure consistency between the kana-kanji conversion candidates and the prediction candidates selected by the user.

[0008]

As described above, the conventional prediction method does not have an efficient mechanism for automatically predicting and presenting an arbitrary part during text input. In view of this point, the present invention converts only a partial character string in front of a prediction start position into a kana-kanji character and adds and presents it before a prediction candidate, thereby presenting a prediction result to a user in an easy-to-understand manner, and performing input operation. It is an object of the present invention to provide a character string predicting apparatus, a character string predicting method, and a storage medium storing a character string predicting program, which can reduce the burden on the user.

[0009]

Therefore, a character string predicting apparatus according to the present invention comprises an input means for inputting character string data, a prediction dictionary storing partial character string data and heading data corresponding thereto, and A partial character string generating unit for generating a partial character string data group for searching the prediction dictionary from the character string data input from the unit, and using the partial character string data group generated by the partial character string generating unit. A dictionary search means for searching the prediction dictionary for heading data corresponding to a partial character string as prediction candidate data; and using partial character string data of the input character string data excluding the character string data used for the dictionary search. Kana-kanji conversion means for performing kana-kanji conversion processing, kana-kanji mixed character string candidate data obtained by this kana-kanji conversion means, and prediction obtained by the dictionary search means It is characterized in that and an output means for outputting to the combined auxiliary data.

According to such a configuration, a partial character string group including the last character is generated from the input character string, and the partial character string group is subjected to a dictionary search as a search character string, whereby prediction can be made from an arbitrary position in the input. It becomes possible. In this case, if an input character string remains before the prediction start position, the display method of candidates becomes problematic, but a partial character string excluding the character string used for the dictionary search in the input character string is converted to a kana-kanji conversion. By adding the character string obtained by Kana-Kanji conversion before the prediction candidate and outputting it, the Kana-Kanji conversion and the input prediction are merged in a natural manner, and the user can feel like a normal Kana-Kanji conversion system. Thus, the input prediction system can be used without discomfort.

The kana-kanji mixed character string candidate data is based on an evaluation value indicating a candidate rank in kana-kanji conversion processing and an evaluation value indicating relevance between the predicted candidate data and the partial character string data. In the case of output by the output means, the output order is determined based on the evaluation value indicating the priority order of the kana-kanji mixed character string candidate data. Kanji conversion candidates can be generated.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a configuration of a character string prediction device according to one embodiment of the present invention. The character string prediction device according to the present embodiment is realized by a computer that reads a program recorded on a storage medium such as a magnetic disk and the operation of which is controlled by the program.

In FIG. 1, an input unit 1 as input means
01 includes, for example, a keyboard, a mouse, and a pen input device, and is used to input a character string or move a cursor;
A command input such as an editing instruction such as insertion / deletion of a character string, an operation instruction such as display of a next candidate, selection of a candidate, etc. is taken in.

After the control unit 102 receives and analyzes the information captured by the input unit 101, it sends necessary information to each processing unit. The processing result of each processing unit is returned to the control unit 102 again, and necessary information is output via the output unit 103 and displayed on, for example, a display.

Here, among various types of information input from the input unit 101, the read character string is sent to the prediction control unit 104 via the control unit 102. The prediction control unit 104 analyzes the received input information, sends necessary information to each processing unit, and returns a prediction candidate to the control unit 102. Next, the control unit 10
2 receives the data related to the prediction candidate selection key from the input unit 101, and sends the input reading character string and the selected prediction candidate to the kana-kanji conversion control unit 114. The kana-kanji conversion control unit 114 analyzes the acquired information, sends necessary information to each processing unit, and outputs kana-kanji conversion candidates to the control unit 10.
Return to 2.

The memory 113 is a storage area for temporarily storing the result of each processing, and is, for example, a RAM. The memory 113 includes a prediction target character string buffer 113a, a search character string buffer 113b, a prediction candidate buffer 113c, a selection candidate buffer 113d, a conversion candidate buffer 113e, and the like.

Here, the prediction target character string buffer 113a
Stores the input character string as a prediction target. This is shown in FIG.

The search character string buffer 113b stores a group of partial character strings for search created by the search character string generation unit 105. Search string buffer 1
13b, as shown in FIG. 12, item 12a indicating a number corresponding to the number of search character strings generated by changing the prediction start position, item 12b indicating the prediction start position for the prediction target character string, and prediction character position It consists of an item 12c indicating a search character string obtained by changing.

The prediction candidate buffer 113c stores a character string obtained as a prediction candidate by the prediction dictionary search unit 106. As illustrated in FIG. 13, the prediction candidate buffer 113 c stores an item 13 a indicating a number corresponding to the number of search results obtained by searching the prediction dictionary 107 and a prediction start position used when searching the prediction dictionary 107. The item 13b includes items 13c, 13d, and 13e corresponding to dictionary numbers, readings, and notations obtained by searching the prediction dictionary 107, respectively.

The selection candidate buffer 113d stores the candidates selected by the user among the prediction candidates. As shown in FIG. 14, the selection candidate buffer 113d stores the prediction candidates selected by the user, and includes, as its items, an item 14a indicating the prediction start position, an item 14b indicating the dictionary number of the prediction dictionary 107, and reading and notation. Items 14c and 14d.

The conversion candidate buffer 113e stores candidates obtained by the kana-kanji conversion unit 108 and evaluation values obtained by the conversion candidate re-evaluation unit 109. As shown in FIG. 15, the conversion candidate buffer 113e is associated with an item 15a indicating the number of candidates, a dictionary number corresponding to the kana-kanji conversion dictionary 110, a reading, and a notation in relation to the candidates obtained by the kana-kanji conversion. Items 15b, 15
c, 15d, item 15 indicating an adjunct word for the candidate
e, for example, an item 15f indicating a conversion evaluation value such as frequency information of the kana-kanji dictionary 110, and a co-occurrence dictionary search unit 1
11, items 15g and 15h indicating the co-occurrence evaluation value and the re-evaluation value obtained by the conversion candidate re-evaluation unit 109, respectively.

Next, processing related to the prediction control unit 104 will be described. The prediction control unit 104 receives a read character string to be predicted via the control unit 102. The received character string is first sent to the search character string generation unit 105. The search character string generation unit 105 generates a search character string group, and returns the search character string group to the prediction control unit 104.

The character string group for search is a set of partial character strings including the last character of the input read character string. For example, when inputting a reading string such as "Kaki no Kai",
In addition to "Kaki no Kagi", a partial character string shortened by one character from the beginning such as "Kino Kagi", "No Kagi", "Kagi", "Kagi" is to be created.

The prediction control unit 104 sends the received search character string group to the prediction dictionary search unit 106. The prediction dictionary search unit 106 searches the prediction dictionary 107 for a heading in which the character string for search and the dictionary reading match forward or completely, and returns the search result to the prediction control unit 104. The prediction control unit 104 returns the received search result to the control unit 102 as a prediction candidate. The prediction candidates returned to the control unit 102 are output to the output unit 103.

When obtaining prediction candidates from search results, it is also possible to calculate an evaluation value of each candidate and assign a priority to the candidates according to the evaluation value.

Next, processing related to the kana-kanji conversion control unit 114 will be described. The kana-kanji conversion control unit 114
For example, when the user selects a prediction candidate presented on the display, the control unit 102 receives the selected prediction candidate and a read target character string (input character string) to be predicted. The read character string from the first character to one character before the prediction start position in the received predicted read character string is sent to the kana-kanji conversion unit 108. The kana-kanji conversion unit 108 refers to the kana-kanji conversion dictionary 110 and returns conversion candidates to the kana-kanji conversion control unit 114. Next, Kana-Kanji conversion control unit 1
14 sends the received conversion candidate and the selected prediction candidate to the conversion candidate re-evaluation unit 109. Conversion candidate reevaluation unit 10
9 sends the received two pieces of information to the co-occurrence dictionary search unit 111, reevaluates the conversion candidates, and returns the result to the kana-kanji conversion control unit 114. The kana-kanji conversion control unit 114 returns the received kana-kanji conversion candidate to the control unit 102. The kana-kanji conversion candidates returned to the control unit 102 are output to the output unit 103
Is output to

The above is the outline of the present system. Next, details of each process will be described using a flowchart.

(A) Processing Operation of Control Unit 102 First, the processing operation of the control unit 102 will be described. FIG.
4 is a flowchart illustrating a processing operation of the control unit 102 according to the first embodiment.

First, in step S201, the memory 113
After initializing the prediction target character string buffer 113a and the selection candidate buffer 113d provided in, the following processing is executed.

That is, in step S202, the input unit 1
The process waits for an input by the user via 01, and if there is an input, the process proceeds to step S203. In step S203, it is determined whether the input is a reading character input, and if so, the process proceeds to step S206; otherwise, the process proceeds to step S204.
Proceed to. To determine whether the input is a phonetic character input,
This can be performed by processing such as referring to a table given in advance. Here, it is assumed that punctuation is not a reading character input. If the input is a so-called Roman character input, buffering is performed here until there is an input corresponding to one character to be read.

In step S206, the input one-character reading is added to the prediction target character string buffer 113a, and the flow advances to step S207. In step S207, it is determined whether the character string to be predicted is two or more characters. If so, the process proceeds to step S208; otherwise, the process returns to step S202, and waits for a user input until the number of characters to be read becomes two or more. In step S208, the information of the prediction target character string buffer 113a is sent to the prediction control unit 104, and the prediction is started.

Here, the process temporarily moves to the prediction control unit 104. Then, the prediction control unit 104 sets the prediction candidate buffer 11
Since the information of 3c is returned to the control unit 102, step S20
In step 9, the information is received and displayed on the input prediction candidate window, and the process returns to step S202 to wait for the next user input. The display processing will be described later in detail.

In step S204, it is determined whether the input is a punctuation mark. If so, step S2
Proceed to step S10, otherwise proceed to step S205. In step S210, the prediction target character string buffer 11
Clear 3a and return to step S202 to wait for the next user input.

In step S205, the data input through the input unit 101 is a key for instructing selection of a prediction candidate.
For example, it is determined whether the key is a prediction candidate selection key. If not, the process proceeds to step S216, where a separately defined process is performed. On the other hand, if so, step S
In step S212, information about the selected prediction candidate is set in the selection candidate buffer 113d.
Proceed to. In step S212, the selection candidate buffer 113
It is determined whether or not the content of the prediction start position 14a described in d indicates the first character. If so, proceed to S215; otherwise, go to step S213.
Proceed to.

In step S213, the information of the prediction target character string buffer 113a and the information of the selection candidate buffer 113d are sent to the kana-kanji conversion control unit 114, and the kana-kanji conversion is started. Since the kana-kanji conversion process is widely known, the description is omitted here. Here, the process temporarily moves to the kana name-kanji conversion control unit 114. Then, since the kana-kanji conversion control unit 114 returns the information of the conversion candidate buffer 113e to the control unit 102, the kana-kanji conversion control unit 114 receives the information in step S214, displays it in the input area, and proceeds to step S215.
In step S215, the information of the selection candidate buffer 113d is displayed in the input area, and the process returns to step S202 to wait for the next user input.

By repeating the above, the control unit 102
The input reading character string information is transferred to the prediction control unit 104, and the input reading character string and selected prediction candidate information are transferred to the kana-kanji conversion control unit 114. The results are received and the next processing is performed. And proceed.

(B) Processing Operation of Prediction Control Unit 104 Next, the processing operation of the prediction control unit 104 will be described.
FIG. 3 is a flowchart showing the processing operation of the prediction control unit 104 in the embodiment.

First, in step S301, the memory 113
After initializing the prediction candidate buffer 113c provided in, the following processing is executed.

That is, in step S302, the control unit 1
From 02, the information of the prediction target character string buffer 113a is received, and the process proceeds to step S303. Step S303
Then, the information of the received prediction target character string buffer 113a is sent to the search character string generation unit 105. Here, once the process
The process proceeds to the search character string generation unit 105. Then, since the search character string generation unit 105 returns the information of the search character string buffer 113b, the information is received in step S304, and the process proceeds to step S305. In step S305, the number of read character strings for search stored in the search character string buffer 113b is set in a variable KR, and the process proceeds to step S306. In step S306, the variable i indicating the current search character string is set to 1, and the process proceeds to step S307.

In step S307, it is determined whether or not i ≦ KR. If i ≦ KR, that is, if there is a search character string for which dictionary search has not been performed, the process proceeds to step S308; otherwise, the process proceeds to step S312.
Proceed to. In step S308, the search character string buffer 11
The i-th read character string of 3b is sent to the prediction dictionary search unit 106. The prediction dictionary search unit 106 returns a search result with reference to the prediction dictionary 107, and receives the result in step S309.

Here, the prediction dictionary 1 used for input prediction
07 will be described. FIG. 8 shows an example of information stored in the prediction dictionary 107. The prediction dictionary 107 includes items 8a, 8b, 8c of "dictionary number", "reading", and "notation".
Consists of For example, in the dictionary number Y01, information such as a reading “Kaki no Gashoku” and a notation (heading) “Summer training camp” is described.

Returning to the description of FIG. 3, step S3 follows.
Go to 10. In step S310, the received search result is set in the prediction candidate buffer 113c, and in step S3
Proceed to 11. In step S311, i is incremented by one, and the process returns to step S307. In step S312, the information of the prediction candidate buffer 113c is sent to the control unit 102, and the process ends.

As described above, a process of creating a partial character string for search from the input character string, searching the prediction dictionary 107, and passing the search result to the control unit 102 as a prediction candidate is performed. The above is the processing flow of the prediction control unit 104.

(C) Processing Operation of Search Character String Generation Unit 105 Next, the processing operation of the search character string generation unit 105 will be described. FIG. 4 shows a search character string generation unit 10 according to the embodiment.
5 is a flowchart illustrating a processing operation of No. 5;

First, in step S401, the memory 113
After initializing the search character string buffer 113b provided in, the following processing is executed.

That is, in step S402, information on the character string buffer 113a to be predicted is received from the prediction control unit 104, and the flow advances to step S403. Step S403
Then, the length of the character string to be predicted is set in a variable L, and the process proceeds to step S404. In step S404, 1 is set to a variable i indicating the prediction start position, and the flow advances to step S405. In order to make the search character string two or more characters, it is determined in step S405 whether or not i ≦ L−1. i ≦
If it is L-1, that is, if the partial character string is two or more characters, the process proceeds to step S406. Otherwise, that is, when the partial character string is less than two characters, step S4
Proceed to 08. In step S406, the character strings from the i-th character to the L-th character and the prediction start position i are set in the search character string buffer 113b, and the process proceeds to step S407. In step S407, i is incremented by one, and the process returns to step S405. In step S408, the information of the search character string buffer 113b is sent to the prediction control unit 104, and the process ends.

As described above, the partial character string group including the last character of the input character string is returned to the prediction control unit 104 together with the prediction start position (start position of the search character string). The above is the processing flow of the search character string generation unit 105.

(D) Processing Operation of Kana-Kanji Conversion Control Unit 114 Next, the processing operation of the kana-kanji conversion control unit 114 will be described. FIG. 7 is a flowchart showing a processing operation of the kana-kanji conversion control unit 114 in the embodiment.

First, in step S701, the conversion candidate buffer 113e provided in the memory 113 is initialized, and the following processing is executed.

That is, in step S702, the control unit 1
02, the information of the prediction target character string buffer 113a and the information of the selection candidate buffer 113d are received.
Go to 03. In step S703, the selection candidate buffer 1
The prediction start position of 13d is set in a variable M, and step S
Proceed to 704. In step S704, the character string from the first character to the (M-1) th character of the prediction target character string buffer is sent to the kana-kanji conversion unit 108. Here, the process temporarily shifts to the kana-kanji conversion unit 108. Then, the kana-kanji conversion unit 108 performs kana-kanji conversion with reference to the kana-kanji conversion dictionary 110,
Since the conversion result is set and returned in the conversion candidate buffer 113e, it is received in step S705.

The kana-kanji conversion method and the dictionary construction method used for kana-kanji conversion are widely known. However, in connection with the co-occurrence dictionary 112 described later, the kana-kanji conversion dictionary used in the embodiment is used. The configuration of 110 will be described.

FIG. 9 shows an example of information stored in the kana-kanji conversion dictionary 110. The kana-kanji conversion dictionary 110 includes items 9a, 9b, 9c, 9d, and 9e of "dictionary number", "reading", "notation", "part of speech", and "frequency". For example, at the position indicated by the dictionary number "K01", information such as "ka" for the reading, "missing" for the notation (heading), "part-of-line verb" for the part of speech, and "2" for the frequency are described. .

Returning to the description of FIG. 7, step S70 follows.
Proceed to 6. In step S706, the conversion candidate buffer 11
The information of 3e and the information of the selection candidate buffer 113d are sent to the conversion candidate reevaluation unit 109. The conversion candidate reevaluation unit 109
The two pieces of information received are sent to the co-occurrence dictionary search unit 111,
Since the information of the conversion candidate buffer 113e to which the reevaluation value is added is returned, the information is received in step S707.
Subsequently, the process proceeds to step S708, and in step S708,
The information of the conversion candidate buffer 113e is returned to the control unit 102, and the process ends.

As described above, the kana-kanji conversion of the character string in front of the prediction start position in the input prediction target character string is performed, and the kana-kanji conversion result is re-used by using the information of the prediction candidate selected by the user. Evaluate and create Kana-Kanji conversion candidates with high accuracy.

(E) Processing Operation of Conversion Candidate Re-Evaluation Unit 109 Next, the processing operation of the conversion candidate re-evaluation unit 109 will be described. FIG. 5 is a flowchart showing the processing operation of the conversion candidate reevaluation unit 109 in the embodiment.

First, in step S501, information of the selection candidate buffer 113d and information of the conversion candidate buffer 113e are received from the kana-kanji conversion control unit 114, and
Proceed to 502. In step S502, the number of candidates stored in the conversion candidate buffer 113e is set in a variable HR, and the process proceeds to step S503. In step S503, the variable i indicating the current conversion candidate is set to 1, and the process proceeds to step S504. In step S504, it is determined whether or not i ≦ HR. When i ≦ HR, that is,
If conversion candidates for which re-evaluation values have not been calculated remain, the process proceeds to step S505; otherwise, the process proceeds to step S510.

In step S505, the ith conversion evaluation value is set in a variable Pt (h), and the flow advances to step S506. In step S506, the i-th conversion candidate and the information of the selection candidate buffer 113d are sent to the co-occurrence dictionary search unit 111. Since the co-occurrence dictionary search unit 111 returns the co-occurrence evaluation value,
In step S507, the information is received, and the variable Pt
(K) is set, and the process proceeds to step S508.

In step S508, Pt (r) is calculated by the following equation (1), and Pt (r) is set in the conversion candidate buffer 113e as reevaluation.

Pt (r) = Pt (h) + 10 × Pt (k) Equation (1) Subsequently, the process proceeds to step S509, i is incremented by 1, and the process returns to step S504.

In step S510, the reevaluation value Pt (r)
Sorts the information of the conversion candidate buffer 113e based on
Proceed to step S511. In step S511, the information in the conversion candidate buffer 113e is
To end the process.

As described above, the kana-kanji conversion candidate is reevaluated using the information on the prediction candidate selected by the user.

(F) Processing Operation of Co-occurrence Dictionary Search Unit 111 Next, the processing operation of the co-occurrence dictionary search unit 111 will be described. FIG. 6 shows a co-occurrence dictionary search unit 11 according to the embodiment.
3 is a flowchart showing a processing operation of No. 1.

First, in step S601, the information of the selection candidate buffer 113d and the i-th information of the conversion candidate buffer 113e are received from the conversion candidate re-evaluation unit 109, and the flow advances to step S602. In step S602, the co-occurrence dictionary 1
The number of co-occurrences registered in No. 12 is set in a variable CR.

Here, the configuration of the co-occurrence dictionary 112 used for the re-evaluation of the conversion candidate will be described. FIG. 10 shows an example of information stored in the co-occurrence dictionary 112. Co-occurrence dictionary 112
Are items 10a, 10b, 10c, 1 of "front dictionary number", "rear dictionary number", "attached word", and "coupling degree".
0d. For example, the first line contains the dictionary number K
Heading 03, ie, “below”, and the heading indicated by dictionary number “Y03” of the prediction dictionary 107, ie, “Please hold a meeting so please join us”, form a co-occurrence relationship via the auxiliary word “no”. This indicates that the co-occurrence degree of coupling is “2”.

Returning to the description of FIG. 6, step S6 follows.
Go to 03. In step S603, 0 is set in a variable Pt (k) as an initial value of the co-occurrence evaluation value, and in step S60
Proceed to 4. In step S604, 1 is set to the number k indicating the current co-occurrence, and the process proceeds to step S605. In step S605, it is determined whether or not k ≦ CR. k
If ≦ CR, that is, if there is any unchecked co-occurrence in the co-occurrence dictionary, step S606
Otherwise, to step S611.

In step S606, the selection candidate buffer 1
It is determined whether the 13d dictionary number matches the k-th rear dictionary number of the co-occurrence dictionary 112. If so, the process proceeds to step S607; otherwise, the process proceeds to step S610. In step S607, it is determined whether the dictionary number of the i-th conversion candidate matches the k-th preceding dictionary number of the co-occurrence dictionary 112. If so, go to step S608; otherwise, go to step S6.
Go to 10. In step S608, it is determined whether the attached word of the i-th conversion candidate matches the k-th attached word of the co-occurrence dictionary 112. If so, step S609
Otherwise, to step S610.
In step S609, the k-th coupling degree is set as a co-occurrence evaluation value in the variable Pt (k), and the flow advances to step S611. In step S611, the co-occurrence evaluation value Pt (k) is returned to the conversion candidate re-evaluation unit 109, and the process ends.

Thus, the received kana-kanji conversion candidate and the selected prediction candidate are collated with the co-occurrence dictionary, and the co-occurrence evaluation value is returned.

Next, an actual processing flow when the present system is used will be described with reference to an example of a screen display shown in FIG.

The readings are input in the order of “ka” and “ki”,
Consider the process at the point when "no" is input. in this case,
Step S203 based on the control of the control unit 102 described above.
Then, the process from step S209 is advanced. in this case,
In the prediction target character string buffer 113a, a read character string “Kakino” is stored as a prediction target character string.
206). Further, since this character string (character string to be predicted) is two or more characters (step S207), the prediction control unit 104
The processing is temporarily shifted to (step S208), and the prediction dictionary 10
7 is performed (contents of the flowchart of FIG. 3).

Further, the prediction control unit 104 shifts the processing to the search character string generation unit 105 (S303, S304), and searches the prediction target character string "Kakino" for the search character string "Kakino".
Receive "kino". These search character strings are stored in the search character string buffer 113b. Further, the process is transferred to the prediction dictionary search unit 106 (steps S308 and S309), and the result of searching the prediction dictionary 107 based on the search character string "Kakino" is received. Here, the results obtained by searching the prediction dictionary are the dictionary numbers Y01 and Y02 of the prediction dictionary 107 in FIG.
The above applies. These search results are stored in the prediction candidate buffer 113c.

Then, the processing in the prediction control section 104 ends, and the processing returns to the processing in the control section 102 again, and the contents of the prediction candidate buffer 113c are displayed via the output section 103 (step S209). Specifically, as shown in FIG. 16, an input prediction candidate window is opened, and two candidates “summer training camp” and “use of firearms” are displayed.

Here, the first candidate “Summer Training Camp” is highlighted as a current candidate. If there is a desired candidate, select it with the cursor. If there is no desired candidate, continue reading and input.

In this case, since there is no desired candidate, if "?" Is continuously input (step S202), it is determined again whether the input data is a read character string (step S203),
The processing after step S206 is performed. again"
At the point of input, the input prediction candidate window disappears as shown in FIG. Further, also when "i" and "gi" are input, the input data is determined to be a read character string in step S203, so that the processing from step S206 is repeated.

Hereinafter, a case where prediction is performed at an arbitrary position in an input character string, which is a feature of the present invention, will be described with reference to a specific example. The processing up to obtaining and displaying the prediction candidate is the same as that described above.

At the point of time when the above-mentioned "Gi" is input, the character string "Kaki no Kaigi" is stored in the prediction target character string buffer 113a as shown in FIG.

In the control unit 102 (the flowchart of FIG. 2), it is determined in step S207 whether or not the character string to be predicted is two or more characters. Since the character string to be predicted is six characters, the process proceeds to step S208. . In step S208, the prediction target character string “Kaki no Kagi” is sent to the prediction control unit 104, and the prediction process starts.

In the prediction control unit 104 (the flowchart of FIG. 3), in step S303, "Kaki no key"
Is sent to the search character string generation unit 105.

Then, in the search character string generation unit 105 (flowchart in FIG. 4), for the character string to be predicted having the length of 6 characters (L is set to 6: step S403), Five search character strings, "Kinokaigi", "Nokaigi", "Kaigi", and "Igi" are generated, and the generated search character string and the number of characters in the prediction target character string Information of a position (prediction start position) as to whether to start from the eye is set in the search character string buffer 113b (step S406). FIG. 12 shows the state of the search character string buffer 113b at this time. Then, the information of the search character string buffer 113b is returned to the prediction control unit 104 (Step S408).

The processing returns to the prediction control section 104 again. In the prediction control unit 104 (the flowchart of FIG. 3), by repeating steps S308 to S311,
All the search character strings stored in the search character string buffer 113b are sent to the prediction dictionary search unit 106, and a heading in which the search character string matches the reading of the prediction dictionary 107 with a forward match or a perfect match is searched for, and is stored in the prediction candidate buffer 113c. Set.

By the processing up to this point, in the prediction candidate buffer 113c, as a candidate whose reading in the prediction dictionary 107 matches the search character string "Kaigi" in front, "A meeting will be held as shown in FIG. Four candidates are stored: "I hold a meeting", "I hold a meeting", "I hold a meeting", and "I hold a meeting".

The processing returns to the control unit 102 (the flowchart of FIG. 2), and the display processing of the prediction candidates is performed in step S209. As a result, the input prediction candidate window opens as shown in Fig. 18, and "We will hold a meeting so please join us""We will hold a meeting""We will hold a meeting"
Four candidates of “I will hold a meeting” are displayed on the display via the output unit 103, for example.

Here, it is assumed that the first candidate is a desired candidate. The user presses a prediction candidate selection key if the candidate is desired. Then, upon receiving the input data from the user, the type of the data is sequentially determined (step S20).
3, S204, S205). In this case, since the input data is the “prediction candidate selection key”, the process proceeds to step S211.
Proceed to.

In step S211, the selected 1
The first prediction candidate “We will hold a meeting so please join us” is set in the selection candidate buffer 113d. FIG. 14 shows the state of the selection candidate buffer 113d at this time.
In step S212, it is determined whether the prediction start position of the candidate stored in the selection candidate buffer 113d is the first character. Since the item 14a of the prediction start position is “4”, the process proceeds to step S213. In step S213, the prediction target character string "Kaki no Kaigi" stored in the prediction target character string buffer 113a (FIG. 11) and the selection candidate "meeting held" in the selection candidate buffer 113d (FIG. 14) are held. Please join us "is sent to the Kana-Kanji conversion control unit 114, and the Kana-Kanji conversion process starts.

In the kana-kanji conversion control unit 114 (the flowchart in FIG. 7), the prediction target character string buffer 113a and the selection candidate buffer 113d are set in step S702.
Receive the information stored in. In this case, since "4" is stored in the item 14a of the prediction start position of the selection candidate buffer 113d, M is set to 4 (step S703). In step S704, the character string from the first character to the third character (value of M-1) of the prediction target character string "Kakinokai", that is, "Kakino" is converted to the kana-kanji conversion unit 108.
Sent to

The Kana-Kanji conversion unit 108 refers to the Kana-Kanji conversion dictionary 110, extracts conversion candidates that match the pronunciation of “Kakino”, and sets the result in the conversion candidate buffer 113e. Conversion candidate buffer 1 at this time
The state of 13e is shown in FIG. Here, the frequency of the kana-kanji conversion dictionary 110 is set in the conversion evaluation value item 15f. At this stage, the items 15g and 15h of the co-occurrence evaluation value and the re-evaluation value are blank.

The processing returns to the kana-kanji conversion control section 114, and then in step S706, the conversion candidate buffer 1
The conversion candidates stored in 13e (FIG. 15A) and the selection candidates stored in the selection candidate buffer 113d (FIG. 14) are sent to the conversion candidate reevaluation unit 109.

In the conversion candidate re-evaluation unit 109 (the flowchart of FIG. 5), steps S505 to S5 are executed.
By repeating step 09, the co-occurrence evaluation value and the re-evaluation value are set in the conversion candidate buffer 113e. Information on the conversion candidate buffer 113e and the selection candidate buffer 113
d (step S501), and in this case,
Since the number of candidates in the conversion candidate buffer 113d is “3”, HR = 3 is set (step S502). Thereafter, the processing from step S504 to step S509 is repeated by the number of candidates. In step S505, the conversion evaluation value “2” of the first candidate in the conversion candidate buffer 113e is set to Pt (h). Step S506
Then, the i-th candidate in the conversion candidate buffer 113e and the information in the selection candidate buffer 113d are sent to the co-occurrence dictionary search unit 111.

The co-occurrence dictionary search unit 111 (the flowchart of FIG. 6) receives the i-th candidate of the conversion candidate buffer 113e and the information of the selection candidate buffer 113d (step S601). A value corresponding to the number of registered co-occurrence dictionaries 112 is set in CR (step S602). Then, the co-occurrence dictionary is searched according to the number of registered cases, and a co-occurrence evaluation value is obtained (from step S605 to step S6).
10).

In this example, the dictionary number 15b of the first (k = 1) candidate of the conversion candidate buffer 113e is K03, the attached word 15e is "no", and the dictionary number 14b of the selection candidate buffer 113d is Y03. The co-occurrence dictionary 112 is searched for a co-occurrence that satisfies the condition that the front dictionary number is K03, the rear dictionary number is Y03, and the attached word is “no”. If it exists, the value of the coupling degree 10d is returned to the conversion candidate re-evaluation unit 109 if it does not exist, and 0 is returned (S611). In this case, since there is a co-occurrence corresponding to the first line of the co-occurrence dictionary 112, “2” is returned as the co-occurrence evaluation value.

The processing returns to the conversion candidate re-evaluation unit 109.
In step S507, the co-occurrence evaluation value is set to Pt (k). In step S508, the reevaluation value Pt (r) is calculated, and the calculated reevaluation value Pt (r) is set in the conversion candidate buffer 113e. In this case, it becomes as follows.

Pt (r) = Pt (h) + 10 × Pt
(K) = 2 + 2 × 10 = 22 When the calculation of the reevaluation values is completed for all the candidates stored in the conversion candidate buffer 113e, step S510
Then, the contents of the conversion candidate buffer 113e are sorted based on the reevaluation value. By the above processing, the contents of the conversion candidate buffer 113e are in the state shown in FIG. The contents of the conversion candidate buffer 113e are sequentially returned to the kana-kanji conversion control unit 114 and the control unit 102 (step S51).
1, S707-S708).

Then, in the control unit 102, step S
At 214, the information stored in the first line of the conversion candidate buffer 113e, that is, the kana-kanji conversion candidate in the front part of the prediction start position is displayed, and then, at step S215, the information stored in the first line of the selection candidate buffer 113d is displayed. Information, that is, prediction candidates are displayed. On the screen, as shown in FIG. 19, the input prediction candidate window disappears, and the prediction candidate “I will hold a meeting so please join us” is inserted in the input area, and the “Kakino” part is “below”. Kana-Kanji characters are converted and displayed.

As described above, only the front part of the prediction start position is converted to Kana-Kanji characters and added to the front of the prediction candidate and displayed. Thus, even if the prediction is automatically performed at an arbitrary position in the text input, There is no need for a separate conversion instruction operation for the processing of the character string in front of the prediction start position, and the user can use this system as if it were a normal Kana-Kanji conversion system.

In the above embodiment, a co-occurrence dictionary consisting of two words is used as shown in FIG.
It is not limited to words, and may be composed of three or more words, and may not be connected. Further, although the coupling degree describes only one value or more, a negative value may be given. That is,
In that case, it is the same as the description of the non-applied or non-priority co-occurrence condition. In other words, if the conditions of the kana-kanji conversion candidate and the selected prediction candidate meet the conditions described in the co-occurrence dictionary, the kana-kanji conversion candidate is rejected or given priority. Is also possible.

In the above embodiment, the expression (1) is used for calculating the re-evaluation value of the kana conversion candidate, but this expression does not limit the present invention.

Further, in the above-described embodiment, the case where the reading character string expressed by the kana is input using a keyboard or the like has been described as an example. However, the form of the input character string is not limited to the kana reading. For example, an input device such as a tablet provided with a character recognition process can directly input kanji. In this case, the input character string is a mixed kanji character string. As described above, a dictionary may be searched using a Kanji Kana character string as a partial character string.

In short, the present invention is not limited to the above-described embodiment, and can be implemented by appropriately changing the scope without changing the gist.

Further, the method described in the above-described embodiment is, for example, a magnetic disk (floppy disk, hard disk, etc.), an optical disk (CD-ROM,
It can be applied to various devices by writing to a recording medium such as a DVD or a semiconductor memory, or can be applied to various devices by transmission via a communication medium. A computer that realizes the present apparatus reads the program recorded on the recording medium, and executes the above-described processing by controlling the operation of the program.

[0100]

As described above, according to the present invention,
By converting only the front part of the prediction start position into Kana-Kanji characters and adding it to the front of the prediction candidate and displaying it, even if the prediction is automatically made at an arbitrary position during text input, it is in front of the prediction start position. No separate conversion instruction operation is required for character string processing, and the user can use this system with the same feeling as a normal Kana-Kanji conversion system.
The efficiency of document input is improved, and Japanese input can be easily performed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of an input prediction device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a processing operation of a control unit in the embodiment.

FIG. 3 is a flowchart showing a processing operation of a prediction control unit according to the embodiment.

FIG. 4 is an exemplary flowchart showing the processing operation of a search character string generation unit in the embodiment.

FIG. 5 is an exemplary flowchart showing the processing operation of a conversion candidate reevaluation unit in the embodiment.

FIG. 6 is an exemplary flowchart showing the processing operation of the co-occurrence dictionary search unit in the embodiment.

FIG. 7 is an exemplary flowchart showing the processing operation of a kana-kanji conversion control unit in the embodiment.

FIG. 8 is an exemplary view showing an example of information stored in a prediction dictionary in the embodiment.

FIG. 9 is an exemplary view showing an example of information stored in a kana-kanji conversion dictionary in the embodiment.

FIG. 10 is an exemplary view showing an example of information stored in a co-occurrence dictionary in the embodiment.

FIG. 11 is an exemplary view showing an example of information stored in a prediction target character string buffer in the embodiment.

FIG. 12 is an exemplary view showing an example of information stored in a search character string buffer in the embodiment.

FIG. 13 is an exemplary view showing an example of information stored in a prediction candidate buffer in the embodiment.

FIG. 14 is an exemplary view showing an example of information stored in a selection candidate buffer in the embodiment.

FIG. 15 is an exemplary view showing an example of information stored in a conversion candidate buffer in the embodiment.

FIG. 16 is an exemplary view showing a screen display when a document is input in the embodiment.

FIG. 17 is an exemplary view showing a screen display when a document is input in the embodiment.

FIG. 18 is an exemplary view showing a screen display at the time of inputting a document in the embodiment.

FIG. 19 is an exemplary view showing a screen display when a document is input in the embodiment.

[Explanation of symbols]

 Reference Signs List 101 input unit 102 control unit 103 output unit 104 prediction control unit 105 search character string generation unit 106 prediction dictionary search unit 107 prediction dictionary 108 kana-kanji conversion unit 109 conversion candidate re-evaluation unit 110 kana-kanji conversion Dictionary 111: Co-occurrence dictionary search unit 112: Co-occurrence dictionary 113: Memory 113a: Prediction target character string buffer 113b: Search character string buffer 113c: Prediction candidate buffer 113d: Selection candidate buffer 113e: Conversion candidate buffer 114: Kana-kanji conversion control unit

Claims (7)

[Claims] An input unit for inputting character string data; a prediction dictionary storing partial character string data and corresponding heading data; and a search for the prediction dictionary based on character string data input from the input unit. Using a partial character string data group generated by the partial character string data generation means, using the partial character string data group generated by the partial character string data generation means to generate heading data corresponding to the partial character string from the prediction dictionary. Dictionary search means for searching as prediction candidate data; kana-kanji conversion means for performing kana-kanji conversion processing using partial character string data of the input character string data excluding the character string data used for the dictionary search; And output means for outputting a combination of the kana-kanji mixed character string candidate data obtained in and the prediction candidate data obtained by the dictionary search means. String prediction apparatus characterized by Bei. 2. The method according to claim 1, wherein the kana-kanji mixed character string candidate data is based on an evaluation value indicating a candidate rank in kana-kanji conversion processing and an evaluation value indicating relevance between the predicted candidate data and the partial character string data. The output means includes an evaluation value indicating the determined priority, and in the case of output by the output means, determines the order of output based on the evaluation value indicating the priority of the kana-kanji mixed character string candidate data. The character string prediction device according to claim 1. 3. A partial character string data group is created from character string data input from an input device, and the partial character string data stored in the storage device and a corresponding heading are created using the partial character string data group. From the prediction dictionary information in which the data is stored, the index data corresponding to the partial character string data is obtained as prediction candidate data, and the partial character strings not used for searching the prediction dictionary among the input character string data are targeted. A character string prediction method comprising: converting kana-kanji characters; and combining kana-kanji mixed character string candidate data obtained by the kana-kanji conversion means with prediction candidate data obtained by searching the prediction dictionary, and outputting the combined data to an output device. . 4. A computer causes a computer to create a partial character string data group from character string data input from an input device, and uses the partial character string data group to store partial character string data stored in a storage device. From the predicted dictionary information in which the corresponding heading data is stored, the heading data corresponding to the partial character string data is obtained as prediction candidate data, and the partial characters of the input character string data that are not used for searching the prediction dictionary A character string prediction for converting a sequence into kana-kanji characters and outputting to the output device a combination of kana-kanji mixed character string candidate data obtained by the kana-kanji conversion means and prediction candidate data obtained by searching the prediction dictionary A storage medium that stores a program. 5. A character string predicting means for generating predicted character string data including partial character string data based on input character string data, wherein the character string predicting means does not use the character string predicting means to generate predicted character string data. Kana-kanji conversion means for converting the remaining partial character string data into kana-kanji character data to generate kana-kanji mixed character string data, and kana-kanji conversion means for predicting character string data generated by the character string prediction means and kana-kanji character string data. A character string prediction device comprising: display means for displaying combined character string data. 6. Based on character string data input from an input device, predictive character string data including the partial character string data is generated, and the remaining partial character string data not used for generating the predicted character string data is generated. A character string prediction method, comprising: converting kana-kanji characters to generate kana-kanji mixed character string data; and outputting a combination of the generated predicted character string data and the generated kana-kanji mixed character string data. 7. A computer, based on character string data input from an input device, generating predicted character string data including the partial character string data, and the remaining partial characters not used for generating the predicted character string data. Stores a character string prediction program for converting column data to Kana-Kanji characters to generate Kana-Kanji mixed character string data, and outputting a combination of the generated predicted character string data and the generated Kana-Kanji mixed character string data. Storage media.