JPH09134351A - Kana/kanji converting method of document processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To execute an operation being suitable to the feeling of a user at the time of correcting and changing a dividing position by permitting grammatical scoring weight to be lightened in a re-converting mode where the dividing position is changeable by means of an operator instruction. SOLUTION: When a KANA character string is inputted from an input part 11 consisting of a keyboard, etc., a control part 14 starts a KANA/KANJI control part 16 in accordance with the input. Then, in a KANA/KANJI converting mode, the KANA input is divided into the plural phrases, scoring is executed by grammatical connecting relation between at least the plural phrases and a KANA/KANJI conversion candidate with a high point is perferably displayed. In the re-converting mode, the deviding position becomes changeable by the operator instruction. Here, at the re-converting mode, the weight of grammatical scoring is made to be light. When the partial decision operation of the KANJI conversion candidate is executed at the time of the re-converting mode, it is especially desirable to make the weight of grammatical scoring light, which is given from the decided part to an undecided part.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a kana-kanji conversion method for a document processing device. In particular, it relates to a kana-kanji conversion method for a Japanese word processor (hereinafter abbreviated as word processor).

[0002]

2. Description of the Related Art In a word processor, a kana key input mode or
Romaji key input mode, keyboard as shown in Figure 1
From (1), when kana characters are entered, they are displayed on the display unit (3). After that, when the "Kanji conversion key" is pressed, it is converted into Kanji (character string mixed with Kanji) as shown in FIG.

Further, when the "Kanji conversion key" is pressed, the next candidate Kanji is displayed as shown in FIG. This display order is
The conversion candidates are displayed in descending order of score. That is,
As a conversion candidate, as shown in Fig. 4, the segment delimiter position (/)
There are various candidate characters. The adjunct word and the inflectional word end are attached to the word. From this combination, choose the one with the highest score.

The general rules for the scoring process are shown below. 1. The score is high because the number of clauses is small (the number of break positions is small). 2. Words that are stored in the Kana-Kanji dictionary in advance and have a high frequency of use have a high score. 3. Learning the kanji actually used, this learned word has a high score. 4. The grammatical connection is judged and the correct score is high.

5. The word stored in the co-occurrence dictionary has a high score. As a result of these scoring, candidates with high scores are preferentially displayed as conversion candidates. By the way, this scoring process is very burdensome. There is no problem if it is used to operate the "Kanji conversion key" after inputting several Kana characters that have been input.

However, if the "Kanji conversion key" is operated after inputting a large number of kana characters that constitute many clauses, the processing analysis time for scoring increases in a geometric progression, It takes a very long time. Therefore, the above rule “1. The number of clauses is small (the number of division positions is small).
Has a high score. ”, The points are scored according to the above-mentioned rules 2 to 5 only for the candidates having a small number of clauses.

In other words, the rule 1 is used to cut off.
Such a truncation will be described below. 6. Those with many phrases are not displayed as candidates. 7. Whether or not the grammatical connection is possible is determined, and the non-acceptance is not displayed as a candidate. 8. Those with low scores are not displayed as candidates.

By the way, in a sentence having a large number of clauses, there are many cases where the delimiter position is different from the user's intention. In this case, the user can change the delimiter position.
Incidentally, regarding such a technique of correcting and changing the division position, Japanese Patent Publication No. 7-7403 (G06F17 / 22), JP-A-4-116764 (G06F15 / 20), JP-A-5-89086 (G06F15) / 20), Japanese Patent Laid-Open No. 6-4519 (G06F15 / 20), etc., and is well known to those skilled in the art.

[0009]

DISCLOSURE OF THE INVENTION The present invention proposes a document processing apparatus which performs an operation suitable for the user's sense when the break position is corrected and changed. At the time of kana-kanji conversion, in order to improve the accuracy, a grammatically correct character string is displayed as described above.

A grammatically incorrect character string has a lower priority or is not displayed. However, there are cases where the language actually used does not follow the grammar. For example, "she doesn't come" is a possible word in reality. However, grammatically, she doesn't come.

[0011] Normally, when "Kanajokonai" is entered and the "Kana-Kanji conversion key" is operated, "She's inside" is displayed and "She does not come" is not displayed. In such a case, it may be corrected in the reconversion mode. Therefore,
The present invention aims to facilitate such modification in a reorganization mode.

[0012]

According to the present invention, kana input is divided into a plurality of phrases, scores are given by at least grammatical connection relations between the plurality of phrases, and a kana-kanji conversion candidate with a high score is given priority. In a kana-kanji conversion method of a document processing apparatus having a kana-kanji conversion mode to be displayed and a re-conversion mode in which the delimiter position can be changed by an operator's instruction, the weight of grammatical scoring in the re-conversion mode is reduced. It is characterized by

Further, according to the present invention, the kana kanji character is divided into a plurality of clauses, the score is given at least by the grammatical connection relation between the plurality of clauses, and the kana-kanji conversion candidate having a high score is preferentially displayed. In the kana-kanji conversion method of the document processing device having a conversion mode and a reconversion mode in which the delimiter position can be changed by an operator's instruction, in the reconversion mode, the delimiter position is changed by an operator's instruction, It is characterized by reducing the weight of the grammatical score given to the undetermined portion by the confirmed portion when the partial confirmation operation of the kanji conversion candidate of the portion is performed.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described with reference to FIG. FIG. 5 shows the operation of this word processor. That is, as in step S2, the "kanji conversion mode" and the "reconversion mode" are determined.

Further, as in step S5, it is divided into a simple "reconversion mode" and a partially confirmed "reconversion mode". The mode is discriminated in this way. Then, as shown in steps S3, S6, and S7, depending on the mode,
The restrictions due to grammatical scoring are removed.

In other words, in the Kanji conversion mode, "Kanojokonai" is "inside the girlfriend" and "She does not come" is not displayed. Further, in the re-conversion mode, “Kanojokonai” is “inside the she”, and “she does not come” is displayed by a plurality of operations of the “Kana-Kanji conversion key”. Further, in the re-conversion mode, if "Kanojo" of "Kanojokonai" is partially confirmed as "she", the remaining "Konajokonai" becomes "Never come".

The second aspect of the present invention will be described with reference to FIGS. 6 to 18.
An embodiment will be described. FIG. 6 is a block diagram of the word processor of this embodiment. (11) is an input unit including a keyboard and a mouse. The input unit (11) is a character key for inputting a kana character string by kana input or roman character input, a kana-kanji conversion key for instructing kana-kanji conversion, a cursor movement key, a temporary confirmation key, a phrase break changing key, and the like. With a key.

(12) is a read-only memory (RO
M). The ROM (12) stores word dictionary data, inflectional / adjunct word data, a connection table, etc., which are necessary when performing kana-kanji conversion. (13) is a readable / writable memory (RAM). RAM
(13) stores a kana input buffer, a candidate table, a phrase candidate table, an output candidate table, a tentative confirmation table, and the like, which are necessary for kana-kanji conversion.

(14) is a control unit. The kana character string and the instruction input from the input unit (11) are given to the control unit (14). The control unit (14) controls the data exchange of each unit of the device according to the given one, and outputs the input character string to the output unit (15) including an LCD display or the like. (16) is a kana-kanji control unit that controls kana-kanji conversion.

The operation of this word processor will be described. The description will be given with reference to the flowcharts of FIGS. 7 and 14, FIGS.
This is performed using the explanatory views of FIGS. When the kana character string "Kanojokonai" is sequentially input from the input unit (11), the control unit (14) activates the Kana-han control unit (16) in response to the input, Run 100.

At step 100, the input key is judged. If the input key is a character, step 110 is executed. If the input key is Kanji conversion, step 130 is executed. If the input key is the next candidate, step 190.
Execute If the input key is tentatively confirmed, step 21
Execute 0.

In step 110, the reading code corresponding to the kana character is stored in the kana input buffer while incrementing the writing position as shown in FIG. In step 120, the temporary confirmation table is initialized. 0 is stored in all the items of the temporary confirmation table. The provisional confirmation table includes the provisionally determined kanji character string codes, the parts of speech, and the number of input kana character strings corresponding to the kanji character strings of the first to eighth phrases.

When the storage is completed, the control is returned to the control unit (4) and the key input is waited. In step 130, the end mark 0 is stored in the input buffer as shown in FIG. In step 140, the conversion start position of the target character string to be subjected to the kana-kanji conversion for the subsequent process is determined from the temporary confirmation table. The provisional fixed table is searched from the first clause to find the total position of the corresponding number of input kana character strings up to a place where the number of corresponding input kana character strings is not zero. If it is not tentatively determined, it becomes 0.

In step 150, the output start clause is determined. The position where the number of corresponding input kana character strings becomes 0 when the temporary fixed table is searched from the first phrase is the output start phrase to the output candidate table for kana-kanji conversion after the next process. In step 160, the pre-connection part-of-speech is determined. When the number of input kana character strings in the first clause of the provisional confirmation table is 0, the part of speech "ALL" that means the beginning of the sentence is set in the pre-connection part of speech.

In other cases, the part-of-speech data stored in the preceding phrase of the phrase in which the number of input kana character strings is 0 in the provisional confirmation table is stored in the pre-connection part-of-speech. Step 17
In 0, the compound phrase kana-kanji conversion process is performed using the word dictionary data as shown in FIG. 9 based on the input buffer, the conversion start position, the output start position, and the pre-connection part-of-speech.

The output of the process is stored in the output candidate table. The contents of this processing will be described later. For example, if there is no provisional confirmation in the state of the input buffer, output data as shown in FIG. 10 is stored. In step 180, when the output start clause is not 0, the data to be stored up to the output start clause in the output candidate table is set as the first data in the temporary confirmation table.
Copy from the clause.

After that, 1 is stored in the display candidate rank and the output unit (1
5) and returns control to the control unit (14). In step 240, the temporary confirmation table is set. Now
An explanation will be given in the case where "Kanajokonai" is input, a Kanji conversion instruction is given, and a temporary confirmation instruction is given immediately after.

It is assumed that for each temporary confirmation instruction, only one clause that is not temporarily confirmed is provisionally assumed from the front. There is no output candidate table immediately after Kana-Kanji conversion.
It has become. The display is "she warehouse" which is first in the output order of FIG. When there is a temporary confirmation instruction, the temporary confirmation table is checked. The provisional confirmation table is searched from the first clause to find the position where the corresponding kana character string is 0.

The data of the corresponding clause of the output candidate table is copied from the data of the display candidate rank of the output candidate table at the position of the checked result. In the case of the above example, since the display candidate rank is 1, the data of the first phrase having the first rank in the output order is copied to the first phrase of the temporary confirmation table. As a result of copying, the temporary confirmation table is as shown in FIG.

In step 190, it is determined whether or not it is immediately after the temporary confirmation in the case of the next candidate. If the result of determination is that it is not immediately after provisional confirmation, step 230 is executed. Immediately after the provisional confirmation, step 200 and subsequent steps are executed. Now, a case will be described in which "Kanajokonai" is input, a Kanji conversion instruction is given, and immediately after that, a temporary confirmation instruction and a next candidate instruction are given.

In this case, step 200 is executed. In step 200, the conversion start position of the target character string for the subsequent kana-kanji conversion is determined from the temporary confirmation table. The provisional fixed table is searched from the first clause to find the total position of the corresponding number of input kana character strings up to a place where the number of corresponding input kana character strings is not zero. In the case of the above example, the conversion start position is 4 from the temporary confirmation table of FIG.

In step 210, the output start clause is determined. The position where the corresponding input kana character string number becomes 0 after searching the temporary fixed table from the first phrase is the output start phrase to the output candidate table for kana-kanji conversion after the next process. In the case of the above example, the output start position is 1 from the temporary confirmation table of FIG. In step 220, the pre-connection part-of-speech is determined. When the number of input kana character strings in the first clause of the provisional confirmation table is 0, the part of speech "ALL" that means the beginning of the sentence is set in the pre-connection part of speech.

In other cases, the part-of-speech data stored in the preceding phrase of the phrase in which the number of input kana character strings is 0 in the provisionally determined table is stored in the pre-connection part-of-speech. In the case of the above example, the pre-connection part-of-speech is “noun” from the provisional confirmation table of FIG. In step 170, the input buffer,
Based on the conversion start position, the output start position, and the pre-connection part-of-speech, the compound phrase kana-kanji conversion process is performed using the word dictionary data as shown in FIG.

The output of the process is stored in the output candidate table. The contents of this processing will be described later. In the case of the above example, FIG.
Output data such as 1 is stored. In step 180, when the output start clause is not 0, the data to be stored up to the output start clause in the output candidate table is copied from the first clause in the temporary confirmation table. In the case of the above example, the first clause of the temporary confirmation table of FIG. 13 is copied to the first clause of the output candidate table, and the data as shown in FIG. 14 is stored.

After that, 1 is stored in the display candidate rank and the output unit (1
5) and returns control to the control unit (14). In step 230, the next candidate is output based on the output candidate table. That is, 1 is added to the value of the display candidate rank, the output unit (15) is instructed to display the contents of the output candidate table of the display candidate rank, and the control is returned to the control unit (14).

The processing of step 170 in FIG. 7 will be described below. The explanation is given by the flowchart of FIG. 14 and FIG.
~ This is performed using the explanatory diagram of Fig. 18. The explanation will be made by using an example of the process of inputting “Kanajokonai” and instructing conversion of Kanji. In step 300, initialization is performed. The output start phrase is stored in the phrase number N, and 1 is stored in the candidate number M (N) of the first phrase. M (N) represents the candidate number of the Nth clause. Store 0 in the number of candidates KOHO that can be combined.

In step 305, the word is searched from the word dictionary data of FIG. 9 based on the input kana character string of FIG. The search result is stored in the candidate table of FIG. What requires the adjunct word processing is whether or not the corresponding kana character string in the input buffer consecutive to the word matches the contents of the inflected word / adjunct word dictionary data using the inflection word / adjunct word dictionary data in FIG. , It is verified whether or not the word matches the part of speech of the inflectional / adjunct word dictionary data, and if matched, it is stored in the candidate table with the inflectional / adjunct word.

In the case of the above example, the result is as shown in FIG. The candidate table consists of the following items. SEQ ID NOs: 1, 2, 3
And the serial number added from the beginning of the table. The starting point is
The beginning of a word is represented by the position of the corresponding kana input buffer. The end point represents the position next to the last position of the word by the position of the corresponding kana input buffer.

In the case of "she" which can be searched from the word dictionary in the above example, the starting point is 0 and the ending point is 4.
The adjunct word length is the number of characters of the inflected word / adjunct word added by the above processing by the inflection word / adjunct word dictionary data. If "not come", it will be 2. Independent word parts of speech,
The part of speech stored in the word dictionary data is stored. The adjunct part-of-speech stores a part-of-speech that represents a backward grammatical connection characteristic when an inflectional ending or an adjunct word is added.

If not added, 0 is stored.
The frequency stores the frequency of the word dictionary data. Shows the frequency of word usage. In step 315, when the phrase number N is 1, the pre-connection part-of-speech and the part-of-speech connection of the Nth clause Mth candidate are tested. A connection table as shown in FIG. 16 is used for the verification. The part-of-speech of the pre-connection part-of-speech is used in the vertical direction, and the N-th clause M-th candidate part-of-speech is used in the horizontal direction. Use this table to test the connection strength. If the combination has a connection strength of 0, the result is NG and step 370 is executed. Otherwise,
Step 320 is executed.

Step 310 is the M-th (N) th section of the N-th clause.
Check candidates. In this check, if N is larger than the output start phrase, the M-th (N-
1) If the end point of the candidate and the start point of the Mth (N) th candidate of the Nth clause are not coincident with each other, the result is NG, and step 350 is performed.
Execute When N is the output start clause, the Mth clause of the Nth clause
(N) If the start point of the candidate is not the conversion start position, NG
Then, step 350 is executed.

In other cases, the result is OK and step 3
Execute 20. In step 350, if the clause number N is larger than the output start clause, step 360 is executed, and if the clause number N is the output start clause, step 410 is executed. In step 360, the phrase number N minus 1 is stored. In addition, a number obtained by adding 1 to the candidate number M (N) of the Nth clause is stored. Then, step 310 is executed.

In step 320, the candidate number M (N + 1) of the (N + 1) th clause is cleared to 0. In step 330, the Mth (N + 1) th candidate of the (N + 1) th clause is checked. The contents are the same as in step 310. If NG, execute step 370. If it is OK, step 330 is executed.

In step 370, the number obtained by adding 1 to the candidate number M (N) of the Nth clause is stored. Then step 31
Execute 0. In step 335, the M-th sentence of the N-th clause
(N) The part-of-speech connection of the (N) candidate and the M (N + 1) th candidate of the (N + 1) th clause is tested. A connection table as shown in FIG. 16 is used for the verification.

The connection table uses the part-of-speech of the preceding clause in the vertical direction, the part-of-speech of an independent word when an adjunct is attached, and the part-of-speech of an independent word when there is no adjunct. In the horizontal direction, the part-of-speech of the after clause and the part-of-speech of the independent word are used. Use this table to test the connection strength. If the combination has a connection strength of 0, the result is NG and step 340 is executed. Otherwise, step 3
Execute 80.

In step 380, the M-th M + 1th clause
It is determined whether the end point of the candidate is the END position of the kana input buffer. If it is END, step 39
Execute 0. Otherwise, execute step 340. In step 390, the combined candidates are evaluated. The evaluation is as follows: 1: The score is high when the number of clauses is small, 2: The score is high when the usage frequency is high, 3: The grammatical connection is high when the connection data in the connection table of FIG. 16 is high. To do so. For example,
If the number of clauses is 1, 100 points, if it is 2, 50 points, etc., may be held in the table.

The sum of these values becomes the evaluation. Step 4
In 00, the candidates that can be combined are stored in the phrase candidate table together with the evaluation in step 390. Number of candidates KO
The value obtained by adding 1 to HO is stored. However, the number of combinations that can be stored in the phrase candidate table is limited to eight.

When the number of candidates KOHO is already 8, the set having the lowest evaluation in the clause candidate table is compared with the evaluation of the candidate to be stored, and the set is stored only when the set to be stored is higher. Store in the position of the group with the lowest evaluation. The number of candidates KOHO in this case is 8
Leave as is. FIG. 18 of the above example is obtained. In step 410, the output candidates are stored from the output start clause of the output candidate table based on the clause candidate table. The output order is determined by a rule such as 1) the order in which the phrase candidate table is highly evaluated, 2: the case where there are homonyms, and the frequency order of the word dictionary data. In the case of the above example, FIG. When KOHO is 0, it is stored in the output start clause of the kana code output candidate table stored in the kana input buffer.

Upon completion of step 410, control returns.

[0050]

According to the present invention, scoring is performed according to the optimum grammar rule according to the mode.

[Brief description of the drawings]

FIG. 1 is a diagram showing a conventional word processor.

FIG. 2 is a diagram for explaining a conventional kana-kanji conversion process of a word processor.

FIG. 3 is a diagram for explaining a conventional kana-kanji conversion process of a word processor.

FIG. 4 is a diagram for explaining segment breaks and conversion candidates.

FIG. 5 is a diagram for explaining the flow of the first embodiment of the present invention.

FIG. 6 is a block diagram of a first embodiment of the present invention.

FIG. 7 is a diagram for explaining the flow of the second embodiment.

FIG. 8 is a diagram for explaining the second embodiment.

FIG. 9 is a diagram for explaining the second embodiment.

FIG. 10 is a diagram for explaining the second embodiment.

FIG. 11 is a diagram for explaining the second embodiment.

FIG. 12 is a diagram for explaining the second embodiment.

FIG. 13 is a diagram for explaining the second embodiment.

FIG. 14 is a diagram for explaining the flow of the second embodiment.

FIG. 15 is a diagram for explaining the second embodiment.

FIG. 16 is a diagram for explaining the second embodiment.

FIG. 17 is a diagram for explaining the second embodiment.

FIG. 18 is a diagram for explaining the second embodiment.

[Explanation of symbols]

 (1) ・ ・ ・ ・ ・ Input section, (2) ・ ・ ・ ・ ・ Main body (Kana-Kanji conversion section), (3) ・ ・ ・ ・ ・ Display section.

Claims (3)

[Claims] 1. A kana-kanji conversion mode in which kana input is divided into a plurality of clauses, a score is given at least by a grammatical connection relation between the plurality of clauses, and a kana-kanji conversion candidate with a high score is preferentially displayed. In a kana-kanji conversion method of a document processing apparatus having a reconversion mode in which the delimiter position can be changed by an operator's instruction, a weight for grammatical scoring in the reconversion mode,
A kana-kanji conversion method, which is lighter than the weight of grammatical scoring in the kana-kanji conversion mode. 2. A kana-kanji conversion mode that divides kana input into a plurality of clauses, scores at least the grammatical connection between the clauses, and preferentially displays kana-kanji conversion candidates with high scores. In the kana-kanji conversion method of a document processing device, when a partial confirmation operation of a kanji conversion candidate is performed in the re-conversion mode, the weight of grammatical scoring exerted by this confirmed portion on an undetermined portion is lightened. Kana-Kanji conversion method characterized by. 3. A kana-kanji conversion mode in which kana input is divided into a plurality of clauses, a score is given at least by a grammatical connection relationship between the clauses, and a kana-kanji conversion candidate with a high score is preferentially displayed. In a kana-kanji conversion method of a document processing apparatus having a re-conversion mode in which the delimiter position can be changed by an operator's instruction, in the re-conversion mode, the delimiter position is changed by the operator's instruction, and the kanji of that part is changed. A kana-kanji conversion method characterized by lightening the weight of grammatical scoring exerted by the fixed part on the unfixed part when the partial fixing operation of the conversion candidate is performed.