JPH0619770B2 - Character processor - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a character processing device for processing Japanese, and more particularly to a character processing device for creating a correct sentence in kana-kanji conversion processing.

[Description of the Prior Art and its Problem] Conventionally, there has been a character processing device that performs kana-kanji conversion using an example when inputting a Japanese sentence. This type of character processing device has an example dictionary that has examples of pairs of adjacent phrases in the phrases that compose a sentence, and is included in the example dictionary in each candidate of the adjacent phrases in the kana-kanji conversion process. It is checked whether or not there is a candidate combination, and if it exists, it is output first.

[Problems to be Solved by the Invention] However, in the character processing device having the above-mentioned conventional example dictionary, the combinations of examples actually used are extremely large, and all the examples should be stored in the example dictionary in advance. Is very uneconomical. To deal with this, there is also a character processing device configured to store only the most frequently used examples in the example dictionary, but the tendency of examples varies depending on the field of text, At the same time, it is known that there is a change in the example, and in such a case, it is not possible to deal with it, and there is a limitation in improving the conversion rate by the example dictionary.

[Means and Actions for Solving the Problems] The character processing device according to the present invention is made to solve the above problems, and includes an input unit for inputting a kana character string, and an input by the input unit. Storage means for storing the kana character string that has been generated, division means for dividing the kana character string stored in the storage means for each clause, and kana character strings for each clause divided by the division means, written in kanji By the conversion means for converting to the bunsetsu candidate group, the independent words of the bunsetsus used adjacently are associated with each other, and the independent dictionary words of the bunsetsus used adjacently are associated with each other. Then, an example storage unit that is newly stored as an example, a candidate display unit that displays one bunsetsu candidate from the bunsetsu candidate group for each bunsetsu, and a display by the candidate display unit, If there is a set of bunsetsu candidates of adjacent bunsetsus in which independent words correspond to examples stored in the example storage unit, the candidate display unit is controlled to preferentially display the candidates forming the set. Display control means, for each of the clauses, a candidate changing means for displaying another one of the clause candidates in the clause candidate group in place of the clause candidate displayed by the candidate displaying means; Selecting means for selecting a desired phrase candidate from the phrase candidates of each phrase displayed by the candidate displaying means, and when the desired phrase candidate is selected by the selecting means,
A storage control unit that associates the independent words of the selected phrase candidates and the phrase candidates selected in adjacent phrases with each other and stores the associated words in the example storage unit as an example, and the display control unit has the phrase candidates. When there is a set of bunsetsu candidates of adjacent bunsetsus, the independent words are stored in the example storage unit if there is a set corresponding to the example stored in the example dictionary unit and the example storage unit. In order to preferentially display the group of candidates corresponding to the example corresponding to the group corresponding to the example stored in the example dictionary means,
It is characterized in that the candidate display means is controlled.

[Embodiment] A character processing apparatus according to an embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a character processing device according to an embodiment.

Reference numeral 1 denotes a CPU that is a microprocessor that controls the entire apparatus, and performs example learning processing described later. Reference numeral 2 is a keyboard controller KBC for controlling the keyboard 3, and 3 is a keyboard KBD made up of a key group such as character keys and function keys. Reference numeral 4 is a CRT controller for controlling the CRT display device 5, and reference numeral 5 is a CRT display device capable of displaying characters such as kanji characters converted into kana-kanji characters. Reference numeral 6 is an FDD controller for controlling a floppy disk device 7 which is an external storage device, and 7 is a floppy disk device FDD which is an external storage device for storing edited document data and the like. Reference numeral 8 is a ROM (read only memory) that stores a control program of the CPU 1 such as an example learning processing program described later, control data, and a dictionary, and 9 is a kana-kanji conversion input buffer 9
a, Kana-Kanji conversion output buffer 9b, reading delimiter buffer 9c, notation delimiter buffer 9d, homophone word buffer 9
9N, and a RAM (random access memory) having a CPU work area and the like. Reference numeral 10 is an example dictionary, and 11 is a rewritable memory for learning examples.

An example learning process of the character processing apparatus having the above configuration will be described with reference to the flowcharts of FIG. 2 and subsequent figures.

First, when a key input is made by the KBD 3, the keyboard KBD input process shown in FIG. 2 is executed. In this process, different processes are performed depending on the input data. First, in the case of inputting character data, step S1
The character data input in step S3 is stored in the kana-kanji conversion input buffer 9a of the RAM 9. In the case of inputting a key for activating the kana-kanji conversion processing, for example, a punctuation mark key or a conversion key, the process proceeds to step 2 and the kana-kanji conversion input buffer 9a.
Kana-Kanji processing is performed on the character data stored in.
This kana-kanji conversion processing is as shown in the flowchart of FIG. That is, in step 2a, the character data string in the kana-kanji conversion input buffer a of the RAM 9 is divided into bunsetsu by the 2-bunsetsu longest matching method, and the delimiter data for reading the kana is read out from the character data of the divided bunsetsu. It is put in the delimiter buffer 9c, and the delimiter data of the conversion result is put in the notation delimiter buffer 9c. The kana-kanji mixed sentence data resulting from the kana-kanji conversion in step 2b is put into the conversion result output buffer 9b. Further, the homophonic words are put in the same homonym buffers 91, 92, 93, ...
Therefore, a plurality of homophones are normally stored in each homophone buffer. Next, in step 2c, it is checked whether or not there is an example between the clauses divided in step 2a. This is a process of checking an example stored in the example dictionary 10 and an example stored in the example learning memory 11 storing an already used example. As a result of checking whether or not there is an example in step 2c, if there is an example, step 2d
Then, the process proceeds to step 2e. In step 2e, it is checked whether or not there are examples in both, and if they are in both, the process proceeds to step 2f. In step 2f, the example learning memory 1
The kana-kanji conversion output buffer 9b is changed so that the example in 1 is output preferentially, and the homophone data in the homophone word buffer is replaced. First, the word found in the example learning memory 11 is changed. Put in the first and shift other homophones in order to lower ranks. This is also the case with homophones that make a pair, that is, the data in the homophone buffers of the adjacent clauses behind. In this case, if the homonym is already in the first rank, it is not necessary to change it. Further, the homophones of the examples in the example dictionary 10 are put in the second rank of the homophone word buffer, and the others are shifted to the third rank or lower. Also in this case, the second
Of course, if the example dictionary data is included in the ranking, it is not necessary to replace it. Further, even when the contents of the example learning memory 11 and the example dictionary 10 match, there is no need to change. When step 2f is completed, the kana-kanji conversion processing is ended. If there is only one in step 2e, the process moves to step 2g, and the kana-kanji conversion output buffer 9 is set so that the found example is preferentially output.
Change b and replace the homophone data in the homophone buffer, put the word found as an example in the first,
Other homophones are sequentially shifted to lower ranks. This is also the case with a homonym paired as an example, that is, a homonym of a phrase adjacent to the back. When step 2g is completed, the kana-kanji conversion processing is terminated.

Returning to the processing of FIG. 2, if the input is made by the cursor movement key, the usual cursor movement processing is performed in step 4. If other keys are input, in step 5, the respective processes corresponding to the key input are performed.

When the homophonic word change instruction key is input, the position of the current cursor is discriminated in step 3 to determine which homophonic word buffer the phrase corresponding to is changed, and the homophonic word is displayed on the CRT display device 5. In the list, the desired homophones are selected from the list. Since the first output homophone is stored in the homophone buffer sequentially, it is possible to know whether or not the selected homophone is the first candidate. In step 6, it is checked whether or not the first candidate homophone has been selected, and in the case of the first rank, the selection processing is ended as it is.
If it is not the first rank, the process moves to step 7, and each pair of the homonym stored in the first rank of the homophone word buffer changed in step 3 and the clauses before and after it is used as the example learning data for the example learning memory. Check whether or not it exists in 11. Since the method of specifically checking the existence of the example is a conventionally known method of matching search, the details thereof will be omitted.
Example learning memory 1 as example learning data in step 7
If it is found that it exists in 1, the process goes to step 8, the example data is deleted, and the process goes to step 11. On the other hand, if it does not exist as the example learning data, the process proceeds to step 9 to check whether the example learning memory 11 has an example learning area. If not, step 10
Then, the oldest example learning data is deleted to create a learning area, and the process proceeds to the next step 11. If there is an example learning area, the process directly moves to step 11 and the changed homophone and the homophone before and after the phrase of the homophone are paired with the outputted homophone, that is, the word of the preceding phrase is changed. The example learning memory 11 stores the resulting homophone pair and the changed homophone and the homophone following the homophone.

In the above explanation, the case where the divided bunsetsu was simply carried out was shown, but in general, the bunsetsu consists of independent words and adjunct words, and the learning data consists of a pair of a front independent part and a rear independent part. Further, the adjunct words of the independent parts are further divided and stored in a form associated with the independent words constituting each independent part. That is, as a simple example, let's do so
In such a case, the case where the conversion result is "operation use" and it is changed to "operation specification" is applicable. In this case, "use" is changed to "specification", and the changed "specification" of the homophone as described above is stored in the example learning memory 11 as a pair of "operation" and "specification". . On the other hand, "I performed a fight"
If you enter, the independent words "specification" and "examination"
The pair has been learned and is further attached to the "specification"
And "wo", which is an annex of "examination", are stored in association with each independent word.

In the above example, the example in the kana-kanji conversion process is shown. However, when the conversion is completed and the document is completed for the time being by detecting the segmentation of the segment and the change of the segmented segment. Can also be implemented in.

[Effects of the Invention] As described above, according to the present invention, in addition to the set of phrases prepared as an example in the example dictionary means in advance, the set of phrase candidates selected by the user requires a special operation. Instead, it is stored in the example storage means as an example, and when displaying the phrase candidates of each clause in the subsequent conversion, it is stored in the example dictionary means or the example storage means in the set of the phrase candidates of the adjacent clauses. If there is a set of bunsetsu candidates, they are preferentially displayed, so that more likely candidates can be obtained based on the example.

In particular, when a set of phrase candidates corresponding to the examples stored in each of the example dictionary unit and the example storage unit exist at the same time, the example stored in the example storage unit that is a set of phrase candidates used by the user Since the set of phrase candidates corresponding to (1) is displayed with priority over the set of phrase candidates corresponding to the examples stored in the example dictionary means in advance, it is possible to display the phrase candidates more suited to the user.

[Brief description of drawings]

FIG. 1 is a block diagram of a character processing device according to an embodiment of the present invention, FIG. 2 is a flowchart showing an example learning process of the present invention, and FIG. 3 is a flowchart showing a kana-kanji conversion process of the present invention. 1 ... CPU, 3 ... KBD, 5 ... CRT, 8 ... R
OM, 9 ... RAM, 10 ... example dictionary, 11 ... example learning memory.

Claims (1)

[Claims] 1. An input means for inputting a kana character string, a storage means for storing the kana character string input by the input means, and a kana character string stored in the storage means is divided for each clause. The dividing means, the converting means for converting the kana character string of each clause divided by the dividing means into a group of candidate clauses in mixed kanji notation, and the independent words of the adjacent clauses used are associated with each other, and the examples are preliminarily used. The example dictionary means stored as, and the example storage means in which the independent words of the clauses used adjacently are associated with each other and are newly stored as an example, and for each clause, one clause from the clause candidate group When displaying by the candidate display means and the candidate display means, there is a set of clause candidates of adjacent clauses in which independent words correspond to the example stored in the example storage means. For example, in order to preferentially display the set of candidates, display control means for controlling the candidate display means, and for each phrase, in place of the phrase candidate displayed by the candidate display means, the phrase candidate A candidate changing unit for displaying another one of the phrase candidates in the group; a selecting unit for selecting a desired phrase candidate from the phrase candidates of each phrase displayed by the candidate changing unit or the candidate displaying unit; When the desired phrase candidate is selected by the means,
A storage control unit that associates the independent words of the selected phrase candidates and the phrase candidates selected in adjacent phrases with each other and stores the associated words in the example storage unit as an example, and the display control unit has the phrase candidates. When there is a set of bunsetsu candidates of adjacent bunsetsus, the independent words are stored in the example storage unit if there is a set corresponding to the example stored in the example dictionary unit and the example storage unit. In order to preferentially display the group of candidates corresponding to the example corresponding to the group corresponding to the example stored in the example dictionary means,
A character processing device for controlling the candidate display means.