JPS62145463A - Kana/kanji (japanese syllabary/chinese character) conversion system - Google Patents

Abstract

PURPOSE:To easily issue the result intended by an operator and to enable an efficient conversion by learning the correction results of a paragraph change operation, a non-conversion key operation or a homonym selection key operation of the operator. CONSTITUTION:A reading of Japanese is inputted to an input processing part 1, a converted KANJI and KANA mixed sentence is corrected by the paragraph change key, the non-conversion key, the homonym selection key and then, a conversion control part 5 registers the contents of the correction in a conversion information management part 3. Thereby, when the same reading is inputted thereafter, according to the previously registered contents of the correction, it is corrected in advance and displayed through an output control part 4. In such a way, the conversion to the intention of the operator can be efficiently carried out while learning.

Description

[Detailed Description of the Invention] Technical Field The present invention relates to a kana-kanji conversion method, and in particular, a learning method for changing the delimiter of units converted by kana-kanji conversion, and a non-conversion key for character string input using the kana-kanji conversion method. The present invention also relates to a kana-kanji conversion method that improves the learning method for homophone selection key operation contents.

Conventional technology In kana-kanji conversion using each unit such as a single kanji, word, or phrase, the delimiter for each unit is inserted using the conversion key J, etc. If the pronunciation of the ``key'' is expressed in kana as is, a common method is to operate the ``non-conversion keys'' individually.

Conventionally, when changing the break, a break display, a break position designating cursor, etc. are used, and it is common to change the break and then select and correct the homophone notation.

The problem here is that if the kana-kanji conversion device divides the input reading sequence into words or phrases with incorrect reading lengths, the operator may somehow correct the error, that is, the reading sequence. It is necessary to correct the division delimiter.

In this case, in the conventional method, no learning is performed regarding the delimiter change operation, so if a similar description is used multiple times in a sentence and it is a conversion display that requires a delimiter change, The input correction operation was extremely troublesome for the operator.

In addition, in the conventional kana-kanji conversion device, the conversion key,
Character strings are input while determining candidates by operating the next candidate key and non-conversion key, and there is no need to learn how to operate the non-conversion key, and not much emphasis is placed on the operation of the non-conversion key.

On the other hand, a mode in which readings are input all at once in advance and homophones selected and non-converted selections are made at once is also quite commonly used in word processors. Also, especially word 2
A kana-to-kanji conversion method that uses ``solid writing'' input is also becoming commonplace, in which characters are input exactly as they are read without being aware of the boundaries between phrases, etc., and the conversion is performed by automatically recognizing punctuation marks and other boundaries.

In such a Kana-Kanji conversion method, 5 Usually, only one copy of words with non-conversion notation is registered in the dictionary, and even if the operator inputs them faithfully to the original text and presses the non-conversion key, the words with homophones and kanji notations are not translated into homophones. In other words, since learning of kana notation is performed only once, if a word written in the same kana notation appears many times in a sentence, it is necessary to perform an input correction operation using the non-conversion key each time, which is extremely difficult. It was bothersome.

Furthermore, there are the following problems with homophone selection. In other words, in kana-kanji conversion for phrases or more, a method is generally used to narrow down conversion candidates by examining the connectivity between words based on word part-of-speech information. in this case,
The method used is to replace the learned candidates along with their part-of-speech information and extract the next candidate. This method generally performed a good analysis when converting phrases, but when converting extended phrases or sentences, if a word candidate in a sentence changes, the structure of the following words will change significantly. This may cause the subsequent word-to-word connection analysis to proceed in an unfavorable direction as conversion candidates.

Purpose The purpose of the present invention is to improve the problems of the conventional kana-kanji conversion method as described above, and to improve the ability of the operator to change the delimiter, change the non-conversion key, or select the homonym selection key each time. The purpose of the present invention is to provide an efficient kana-kanji conversion method that facilitates the production of conversion results.

Structure In order to achieve the above object, the kana-kanji conversion method of the present invention inputs the Japanese pronunciation and converts it into a kanji-kana mixed sentence by searching a word dictionary. Equipped with a delimiter change key for the user to modify unit delimiters, a no-conversion key for the user to modify the converted notation to kana notation, and a homophone selection key for modifying the converted notation to homophone notation. an input means, a word delimiter change registration means for storing the operation contents modified by the delimiter change key, a non-conversion modification registration means for storing the operation contents modified by the no-conversion key, and a word delimiter change registration means for storing the operation contents modified by the no-conversion key; and homophone selection correction registration means for storing the operation contents that have been changed, and when the same reading is input after the correction contents have been registered in each of the registration means, the conversion display is
A feature is that the display is modified in advance according to the modification operation details registered in the registration means.

Hereinafter, the configuration of the present invention will be explained in detail by way of examples.

FIG. 1 is a processing flowchart of a kana-kanji conversion method showing an embodiment of the present invention. In the figure, 10 is a word dictionary, 11 is a grammar dictionary, 12 is a word break change registration table, and 13
14 shows a non-conversion correction registration table, 14 shows a homophone selection correction registration table, and 101 to 124 show processing steps. The structure of the dictionary and table will be explained in detail later.

Note that in FIG. 1, rKKCJ is an abbreviation for kana-kanji conversion.

An outline of the processing of this embodiment is as follows.

Note that the numbers in parentheses after the processing description indicate the numbers of the processing steps.

When the initial processing (101) is completed, an input request is displayed on the CRT (102), and if the operator has the key power (1
03), check whether it is a character input or other input (10
4) In the latter case, is it a modification operation of the conversion display?
Others (e.g. other "word processor features",
or other "conversion functions") (112, 114
).

If it is a termination operation or other operation, the corresponding processing is performed (113, 115), and if it is a modification operation, modification processing is performed (116). If there is a change that separates
The word delimiter change registration process stores the operation details in main memory.
are also stored in the form of word break change registration table 12, which will be described later (
117, 118).

If the modification process is without conversion, the operation contents are stored in the main memory in the form of the non-conversion modification registration table 13 by a non-conversion modification registration process (119.degree. 120).

In addition, when the correction process is homophone selection, the operation contents are stored in the main memory in the form of the homophone selection correction registration table 14 by the homophone selection correction registration process (121, 122
).

Then, when the word break change registration, non-conversion correction registration, or homophone selection correction registration processing is performed, the correction display is edited accordingly (123), and the edited result after correction is displayed on the CRT (124). .

When the operator's input is a character string, after preprocessing the character string (105), the word dictionary 10 and the grammar dictionary 1 are used.
1 to extract a word candidate string (106).

When a word candidate string is selected, first, the previously registered word break change registration table 1 is selected by priority processing for candidates other than break change changes.
2, and if there is a corresponding one, change the first conversion candidate to that candidate (107).

Next, refer to the previously registered non-conversion correction registration table 13 for priority non-conversion candidates, and if there is a corresponding one, change the notation of the first conversion candidate to the literal kana notation (1
08).

Also, for those that prioritize homophone selection candidates.

Refer to the previously registered homophone selection correction registration table 14, and if there is a corresponding one, change the first conversion candidate to that candidate (
109), performs the next converted character string editing process, and displays it on the CRT as a conversion candidate (110°111).

As mentioned above, in the kana-kanji conversion method shown in FIG. 1, the operator's word break change operation.

By registering and learning the contents of the non-conversion correction operation and the homophone selection correction operation, the results can be used preferentially to speed up the conversion process.

FIG. 2(A) is an explanatory diagram showing an example of a modification operation related to word break change among the modification operations in FIG. 1.

Here, when the "yomi" is input for the input "original text" and the "converted sentence" is displayed, the operator can change the notation that is different from the original text using the delimiter change key and homophone selection key (conversion key). Modify as shown.

The modification operation is, for example, "Fukushiya" → "F</shiya"...Example to shorten it: "Shiya J+r Chosa" → "Shiyachi Yosa" → "President,
``Sa'' ・・・Example of lengthening ``Sa'' Ju ``ma'' → ``sama'' → ``sama'' ・・・・・・It is done as in the example of lengthening.

FIG. 2(B) is an explanatory diagram showing an example of a modification operation related to no-conversion selection among the modification operations in FIG. 1.

Here, the "reading" of the input "original text" is input, and when the "converted sentence" is displayed, the operator corrects the notation that differs from the original text using the non-conversion key as shown in the example. Modification operations are performed, for example, as follows: "Half speed" → "Issado", "Go" → "Go", "Makoto" → "Makoto".

FIG. 2(C) is an explanatory diagram showing an example of a modification operation related to homophone selection among the modification operations in FIG. 1.

Here, the "yomi" is input for the input "original text", and when the "converted sentence" is displayed, the operator corrects the notation that differs from the original text using the homophone selection key (conversion key) as shown in the example. . For example, the correction operation is performed as follows: ``Background'' → ``Dear Sir,''``Reporter,'' ``Your company,''``Seongei'' → ``Seongei.''

FIG. 3(A) shows the case where the correction operation as shown in FIG. 2(A) has been performed. FIG. 3 (B), which is a diagram illustrating the contents of the word break change registration table 12, shows the contents of the non-conversion correction registration table 13 when the correction operation as shown in FIG. 2 (B) is performed. Illustrative diagrams and Figure 3 (C)
2 is a diagram illustrating the contents of the homophone selection correction registration table 14 when the correction operation shown in FIG. 2(C) is performed.

The word break change registration table 12 shown in FIG. 3(A) shows (a) the reading length before the change, (b) the reading length after the change, of the part that has been changed using the 1 break change key. (C) Change type, (d) Change part reading character string, and (8) Part of speech of the previous word are stored.

In addition, the change type is II if the original word is lengthened.
If I II is shortened, register It 217. In addition, the reading length after the change is the reading length of the word after the combination (back) in case of change type 1, and the reading length of the first divided word in case of change type 2. be.

The non-conversion correction registration table 13 shown in FIG.
b) Reading character string + (c) W record flag and notes are stored. In addition, the homophone selection correction registration table 14 shown in FIG. 3(C) includes (a) reading length;

(b) A reading character string, (C) a notation character string, and (d) a registration flag and notes are stored.

In addition, the registration flag in the non-conversion correction registration table 13 and the homophone selection correction registration table 14 is
11), l#O'g is displayed when the registration is unregistered, the registration has been deleted, or the registration has become invalid.

When registering only the part indicated by A in the word break change registration table 12 shown in FIG. 3(A), only the part indicated by A in the non-conversion correction registration table 13 shown in FIG. 3(B) is registered. In this case, even if only the part shown in B is registered,
Full effects can be expected.

Similarly, the homophone selection correction registration table 1 shown in Figure 3(C)
4, sufficient effects can be expected even if only the part indicated by A is registered.

FIG. 4 is a block diagram of a character string input device using a kana-kanji conversion method showing an embodiment of the present invention. In the figure, 1 is an input preprocessing unit, 2 is a candidate extraction unit, 3 is a conversion information management unit, and 4
5 indicates an output control section, and 5 indicates a conversion control section.

When the reading is input, the input preprocessing unit 1 separates the "solid writing" sentences and classifies them into words and adjuncts, or Performs processing such as searching for frequently occurring kana sentences and excluding them from conversion targets.

Next, the candidate extraction section 2 converts the corresponding word into kanji under the control of the conversion control section 5, and transfers the converted information or excluded information to the conversion information management section 3 for storage management. After being edited by the conversion control unit 5, it is output to the output control unit 4, and after performing return processing such as backtracking, a sentence containing kanji and kana is output.

FIG. 5 is a perspective view of a specific example of the character string input device shown in FIG. 4. As shown in FIG. 5, the input device is divided into a display section 51 and a keyboard section 52, and the display section 51 has a text area 53. A display area 54 for conversion display lines, correction lines, reading input, etc. is included.

The keyboard section 52 also includes 1 function key 55, character key 56. Numeric keypad 57. Conversion control keys and others are provided. Modification keys for the above and others.

It includes a candidate selection key, a delimiter change key, etc.

FIG. 6 is a diagram showing a detailed configuration of the candidate extracting section of FIG. 4.

The candidate extraction section 2 includes a candidate extraction control section 6. Candidate word string storage unit 7. It is composed of a dictionary search section 8 and a conversion dictionary storage section 9.

When a kana character string signal is input to the candidate extraction control unit 6, the control unit 6 searches the conversion dictionary storage unit 9 from the beginning using the dictionary search unit 8, and extracts all conversion candidates corresponding to the reading. . When there is no conversion candidate, the dictionary search section 8 notifies the candidate extraction control section 6 of this fact. The candidate word string is stored in the candidate word string storage section 7.

FIG. 7 is a processing brochure of a registration method when changing and correcting word breaks according to the present invention.

In this process, when there is a correction request from the operator, it is determined whether the registration conditions are satisfied or not (201), and if they are, the registration position of the registration table is set at the 1st f#.
02), Register the reading length before changing the delimiter (203
), the second step is to register the reading length after changing the delimiter.2
04), please register the type of break change in the 3rd tr20
5), fourthly, register the changed part reading character string 206)
, and finally register the part of speech located before the change part (
207).

Note that if some conditions are omitted from the registration table 12 (see A in FIG. 3(A)), the above process 207 is unnecessary.

The registration condition for the above process 201 is the correction mode.

Also, when using normal kana skills, refer to a dictionary and think of possible word breaks. Furthermore, as in the example of "Shiya" in the case of a correction operation, a partial string of words remaining from a previous delimiter change operation is not registered if the delimiter is changed again.

The registration position setting in the process 202 is performed as shown in FIG. 3(A), for example.
As in the example shown in , a fixed-length table is set up and registered in the order of serial numbers. When the table becomes full, it returns to the beginning of the serial number and overwrites and registers one after another.

As described above, registration proceeds in order from the top, but reference is made from the latest registration position and reference is made in reverse order of serial numbers (hereinafter referred to as "ring method").

FIG. 8 is a processing flowchart of the non-delimiter change candidate priority processing (see 107 in FIG. 1) according to the present invention.

Here, a candidate other than the delimiter change means a candidate different from the display candidate before the delimiter change at the time of registration. For example, in the example shown in Figure 2 (A), the input reading is "Fukushiya...
"...", the candidate is "Fukushiya".

There are "fu<L/J, rfu<" and "fu", but as a conversion candidate, "fukushi" is not followed by "ya" in the later word, and is "fu".
It doesn't follow "Mokushiya", so "Fukuja" and "Fuku"
is selected.

In general, words with long pronunciations are evaluated highly by the device, so "Fukushiya" is selected as the first candidate, and the next candidate with a different length is "Fuku". Therefore, "Fukushiya...
In the example of ``...'', if there is a registration as shown in FIG. 3(A), a candidate with the pronunciation of ``Fuku'' is selected from the conversion candidate string.

Returning to the process shown in FIG. 8, the explanation will be continued. This process is
After extracting the word candidate string, it is determined whether the reference conditions of the registration table are met (301), and if they are, the word break change registration table 12 is searched (302). And J
It is determined whether or not there is a suitable candidate (303), and if there is, the first candidate for conversion display is corrected to the corresponding candidate (304).

The reference conditions for the above process 301 are input mode and normal kana-kanji conversion.

FIG. 9 is a processing flowchart 1 of the registration search shown in FIG.

When searching the registration table 12 using the non-word boundary candidate priority process, first set the reference position to the latest registered position.
01), determine whether the starting part of speech of the previous word matches or not 402), and if they do not match, determine whether there is still something to refer to 406), then move the reference position to the previous position. The latest registered position is set (407).

If the part of speech of the previous word matches, then it is determined whether the pronunciation matches the conditions at the time of registration (403). If they do not match, it is determined whether there is anything else to refer to, and the reference position is set to the latest registered position immediately before (40
6,407).

When a match occurs, it is determined whether or not it is suitable as a candidate outside of delimiter change (404), and the position of the corresponding candidate is set (404).
05). If the candidate is not suitable as a candidate for delimiter change, it is determined whether there is anything else to refer to, and if there is no other candidate, it is set to "No corresponding candidate" (408).

In this process flowchart, if the part of speech is not registered in the registration table 12, process 402 is deleted.

FIG. 10 is a processing flowchart for comparing readings in FIG. 9.

In process 403 in FIG. 9, it is determined whether the pronunciation matches the conditions at the time of registration. This is to find a match with the pronunciation in the registration table 12 from the break position after the word. For example, the registration example is shown in FIG. 3(A).

In "Fukusha", the input is "...Fukushiyagi..."
・”.

In cases such as "...Fukushachiyoga...", the underlined part is the part that has the same reading.

Therefore, it is necessary to determine whether or not it matches the reading character string part before the change 501), and if it matches, it is determined whether the change type is 1 or not (502), and if it is 1, the remaining reading character string after the change. It is determined whether or not it matches the portion (503). If they do not match, it is set that the conditions at the time of reading registration do not match (505). Further, if they match, the reading is set to match the conditions at the time of registration (504).

FIG. 11 is a processing flowchart 1 of the delimiter change non-candidate search process in FIG. 9.

In this process, for conversion candidates whose delimiters have been changed by the operator, at least the conversion display in the conversion table does not match the display intended by the operator. If there are any candidates, they will be displayed with priority.

As described above, the different candidates have different reading lengths according to the evaluation of the candidates determined by the conversion device, and the one in the next rank is selected. If there are no candidates with different reading lengths, the device performs the candidate word string extraction process (1) shown in Figure 1.
Display the first candidate determined in step 06).

To determine whether or not the candidate is suitable as a candidate other than the delimiter change shown in FIG. 11, first set the candidate comparison position to the light head position of the candidate column table.
01), determine whether it matches the reading length before the change in the registration table 12 (602), and if there is a match and there is still something to compare, set the next comparison position to the next position in the candidate sequence table. and repeats the process (602, 606, 607).

If it does not match the reading length, it is determined whether the change type of the registration table is 1 (603), and if it is 1, the candidate column table position is set as a candidate for delimiter change (605). Also, if the change type is not 1, it is determined whether it matches the length of the reading after the change in the registration table (604), and if it does not match, it is determined whether there is still something to compare, and that If so, it is sequentially set to the next position (606, 607).

If there is nothing to compare, "Not Applicable" is set as a candidate for delimiter change (608). Furthermore, here, when the change type is type 2 (not 1), the delimiter non-change candidates are limited to those that match the reading length after the change, but this part may be omitted.

FIG. 12 is a processing flowchart of a registration method for non-conversion modification according to the present invention.

In this process, when there is a correction request from the operator, it is determined whether the registration conditions are met (251), and if they are, the registration position in the registration table is set as the first step (251).
2) Second, register the reading length (253). Thirdly, the reading character string is registered (254), and fourthly, the registration flag is set to "register" (255).

Note that when some conditions are omitted in the registration table 13 and only the pronunciation character strings are registered (see A in FIG. 3(B)), the above processes 253 and 255 are unnecessary.

Similarly, if you want to register only the reading length and reading string (
(See B in Figure 3 (B)) above)! 11255 becomes unnecessary.

The registration conditions for the process 251 are as follows: when the user is in the correction mode and is using normal kana skills, the dictionary is referred to and candidate word delimiters are considered. Also, the registration location setting can be done in the third
As shown in the example shown in Figure (B), when the 0 table in which a fixed-length table is set and the serial numbers are registered in the order of serial numbers is full, the sequence returns to the beginning of the serial numbers and is overwritten and registered in sequence ( ring method).

FIG. 13 is a diagram of the non-conversion candidate priority processing (see FIG. 1 108) according to the present invention.

In this process, after extracting a word candidate string, it is determined whether the reference conditions of the registration table are satisfied or not (3-51).

If it is complete, a search for the registration table (352) is performed.

Then, decide whether or not there is something that applies.353)
, in some cases, change the notation of the first candidate for conversion display to kana (
354).

It should be noted that the reference condition for the process 351 is an input mode and normal kana-kanji conversion.

Figure 14 shows the non-conversion correction registration table 13 search (3
52) is a processing flowchart.

When searching the registration table using non-conversion candidate priority processing,
First, the reference position is set to the latest registered position (451), and it is determined whether the reading lengths match (452). If they do not match, it is determined whether there is still something to refer to. 456), if there is, the reference position is set to the previous latest registered position (457). If not, a state of no corresponding non-conversion candidate is set (458).

If the reading lengths match, then it is determined whether the reading character strings match (453). Also.

If they do not match, it is determined whether there is anything else to refer to (456), and the reference position is set to the previous position (457). If the reading lengths above match, check whether the registration flag is in the "registered" state (
454), if the status is "registered", a status is set that there is a corresponding non-conversion candidate (455).

If there is nothing else to refer to, a state of no corresponding non-conversion candidate is set (458). Furthermore, even when the registration flag is not in the "registered" state, a state of no non-conversion candidates is set (458). Note that in the case where the reading length is not registered in the registration table, processing speed can be given priority by storing the reading length as well, if there is sufficient storage space.

FIG. 15 is a processing brochure of the registration method when selecting homophone correction according to the present invention.

In this process, when there is a correction request from the operator, it is determined whether the registration conditions are met (551), and if they are, the registration position in the registration table is set first (552), and the reading length is is registered (553).

Second, the reading character string is registered 554), the third day's notation character string is registered 555), and fourth, the registration flag is set to "register" (556).

Note that if some conditions are omitted from the registration table 14 (see FIG. 3(C)), the above processes 503 and 506 are deleted. As mentioned above, even if it is deleted, there is a considerable effect.

The registration condition for process 551 is that when the user is in the correction mode and the normal kana manual is used, a dictionary is referred to and candidate word delimiters are considered.

Further, to set the registration position, for example, a fixed length table as shown in FIG. 3(C) is set, and registration is performed in order of serial number. When the table becomes full, the serial numbers are overwritten and registered again from the beginning. Then, the reference movement is performed from the ascending Bm new position,
Reference is made in the reverse order of serial numbers (the ring method).

FIG. 16 is a processing flowchart of the homophone selection candidate priority processing (see FIG. 1 109) according to the present invention.

In this process, the word candidate comparison position is set to the beginning position of the candidate array (651), and it is determined whether the reference conditions of the registration table are satisfied (652). If it is in place,
After searching the registration table 13 (653), it is determined whether a corresponding item exists (654). If there is a corresponding one, the notation of the first candidate for conversion display is set as the notation of the corresponding candidate (655).

On the other hand, if the reference conditions for the registration table are not in place.

Alternatively, if there is no matching item in the registration table, it is determined whether there are any more candidates for comparison (656).

If not, the process ends, and if there is, the candidate comparison position is set to the next position in the candidate array (657). Note that the reference condition for processing 652 is an input mode and normal kana-kanji conversion.

FIG. 17 is a processing flowchart of the registration table search (653) in FIG. 16.

When searching the registration table 14 with priority given to homophone selection candidates, this process should set the reference position to the latest registered position 7-5
1), determine whether the reading lengths match or not 752)
, if there is a mismatch, it is determined whether there is still something to refer to (757), and the reference position is set to the previous latest registered position (758).

If the reading lengths match, it is determined whether or not the reading character strings match (753). If it doesn't match,
Determine whether there is anything else to refer to (757L1
A reference position is set to the previous position (758). If they match, it is determined whether the registration flag is in the "registered" state (754), and if it is in the "rregistered" state, it is determined whether the written character strings match (755).

If the written characters match, the table position where the corresponding homophone candidate is located is set (756), and the registration flag is set to "register".
If there is no corresponding homophone candidate, or if the 1-notation character strings do not match, set the state of no corresponding homophone candidate (75
9). Note that if the registration of reading length and registration flag is omitted in the registration table 14, processes 752 and 754 are deleted.

In the above processing, checking whether or not the reading length is applicable is effective when there is sufficient space in the main memory and priority is given to processing speed.

In addition, the registration flag is kept in the "registered" state at the time of registration, and if there is a reconversion with the same reading, the registration flag is changed to "cancelled" and the registration candidate is invalidated.
If a homophone with the same reading is selected, it will be registered in a new position.

FIG. 18 is a processing flowchart showing another embodiment of the present invention.

In this embodiment, in addition to the content shown in FIG. 1, when there is a conversion of a word unit having the same pronunciation as the stored correction content as described above, the registered content is invalidated. In other words, if there is a no-conversion operation or a homophone selection operation during correction, first the non-conversion correction registration table 13 and the non-conversion correction registration table 13 are The homophone selection correction registration table 14 is forged, and if there are words registered with the same pronunciation, those registrations are canceled. After that, non-conversion correction registration or homophone selection correction registration is performed.

FIG. 19 is a flowchart of non-conversion modification registration cancellation processing according to the present invention.

In FIG. 19, as in the case of non-conversion registration, it is first determined whether the reference conditions for the registration table are met (801) L
, then set the reference position to the latest registered position (802)
. Then, it is determined whether the reading lengths match as a comparison condition (803), and if they match, it is determined whether the reading character strings match (804).

If both match, "cancel" the registration flag of the reference position.
(805). Also, if the reading length or reading character strings do not match, it is determined whether there is still something to refer to (806), and if there is, the reference position is changed to 1.
The previous g is set to the latest recording position (807), and the process moves to repeated comparison processing (803-804).

When there is nothing to refer to, the process ends.

FIG. 20 is a processing brochure for homophone selection, modification, and deregistration according to the present invention.

First, it is determined whether the reference conditions for the registration table are in place (901), and if so, the reference position is set to the latest registered position (902).Next, whether the reading lengths match or not is determined. If they match, it is determined whether the pronunciation character strings match (903), and if both match, the registration flag of the reference position is set to "cancel" (903). 905).

In addition, if the reading length or reading character string does not match, check whether there is anything else to refer to (906).
), if there is, the reference position is set to the latest registered position immediately before (907), and the process moves on to comparing and determining the reading length again (903). If there is nothing to refer to, the process ends.

According to this embodiment, consistency with other conversion operations is performed,
This has the effect that no defects occur.

FIG. 21 is a processing broach yard showing still another embodiment of the present invention.

In this embodiment, in addition to the contents shown in FIG. 1, the contents of the non-conversion operation and the homophone selection operation at the time of correction are stored in the order of occurrence. FIG. 21 is an example of this, in which the non-conversion correction registration table 13 and the homophone selection correction registration table 14 shown in FIG. 1 are combined to form a non-conversion/homophone correction registration table 15. , No conversion/homophone selection correction registration process (127)
Then, non-conversion/homophone selection candidate priority processing (128) is performed.

In this embodiment, non-conversion processing is regarded as a type of homophone selection and treated in the same way in chronological order. This has the advantage of simplifying the configuration of the registration table and improving processing speed. be.

An example of the correction operation in the above embodiment is shown in FIG. 22, and an example of the contents of the non-conversion/homophone selection correction registration table 15 is shown in FIG.

In each of the above embodiments, an example was shown in which the contents of the modification operation were stored in the main memory, but this is not the case in the external storage device i!
! It goes without saying that the information may be stored in a disk drive (for example, a disk device, etc.).

Effects As described above, according to the present invention, by learning the correction results caused by the operator's delimiter change operation, non-conversion key operation, or homophone selection key operation, it becomes easier to obtain the conversion result that the operator intends, increasing efficiency. This method has the excellent effect of realizing a kana-kanji conversion system.

[Brief explanation of drawings]

FIG. 1 is a processing flowchart of a kana-kanji conversion method showing an embodiment of the present invention, FIGS. 2(A) to (C) are explanatory diagrams showing examples of change and correction operations in FIG. A
) is a diagram showing an example of a word break change registration table, and Figure 3 (B
) is a diagram showing an example of a non-conversion correction registration table, FIG. 3(C) is a diagram showing an example of a homophone selection correction registration table, and FIG. 4 is a kana-kanji conversion method character string showing an embodiment of the present invention. A block diagram of the input device, FIG. 5 is a perspective view of the input device of FIG. 4, FIG. 6 is a block diagram of the candidate extraction unit of FIG. 4, and FIG. FIG. 8 is a processing flowchart for giving priority to candidates other than word break changes. Figure 9 is a flowchart of the word break change registration table search process, Figure 10 is a flowchart of the reading comparison process, and
The figure is a processing flowchart of candidate a search outside of delimiter change in Fig. 9, Fig. 12 is a flowchart of registration processing at the time of non-conversion modification, Fig. 13 is a processing flowchart of non-conversion candidate priority, and Fig. 14 is non-conversion correction registration. Table search processing flowchart 1. The 15th is a flowchart of the registration process when selecting and correcting homophones. Figure 16 is the flowchart of processing that prioritizes homophone selection candidates. Figure 17 is the process flowchart for homophone selection and correction '18 table search. Processing flowchart, FIG. 18 is flowchart 1 of kana-kanji conversion processing showing another embodiment of the present invention, FIG. 19 is a processing flowchart of non-conversion correction registration cancellation, and FIG. 20 is homophone selection selection correction registration cancellation. processing flowchart,
Fig. 21 is a processing broach yard showing still another embodiment of the present invention, Fig. 22 is a diagram showing an example of a correction operation in the embodiment shown in Fig. 21, and Fig. 23 is a non-conversion/homophone selection correction. FIG. 3 is an explanatory diagram of the contents of a registration table. Input pre-processing unit, 2: Candidate extraction unit, 3: Conversion information management unit, 4: Output control unit, 5: Conversion control unit, 6: Candidate extraction control unit, 7: Candidate word string storage unit, 8: Dictionary search Section 9: Conversion dictionary storage section. Patent applicant Ricoh Co., Ltd. - Figure 1 Figure 3 B (C)
14? Figure 10 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 Figure 14 Figure 17 Figure 19 Figure 20 Figure 21 Jll) January 31, 1985 (, 1960 Patent Application No. 288187 2, Title of invention: Kana-kanji conversion method 3, Relationship with the person making the amendment case: Patent applicant 4, Attorney 17゛7, Base Correction details i ji4[j,,i,,',
[7Q,) to (C)・Figure 23 to Figure 3

Claims (5)

[Claims] (1) In the kana-kanji conversion method, which converts the Japanese pronunciation into a kanji-kana-mixed sentence by searching a word dictionary, the user can use the delimiter change key to modify the delimiter of the converted units. , an input means having a no-conversion key for the user to modify the converted notation to kana notation, and a homophone selection key for changing to homophone notation, and storing the operation contents modified by the delimiter change key. a word break change registration means for storing the operation contents modified by the no-conversion key, a non-conversion modification registration means for storing the operation contents modified by the homophone selection key, and a homophone selection modification registration means for storing the operation contents modified by the homophone selection key. and when the same reading is input after the correction content is registered in each of the registration means, the conversion display is
A Kana-Kanji conversion method characterized in that the Kana-Kanji conversion method is characterized in that the Kana-Kanji conversion method is corrected in advance and displayed according to the correction operation details registered in the registration means. (2) When the change display is divided into words, the word division change registration means stores the pronunciation before the change when a change operation is performed in the word unit, and thereafter stores the pronunciation of the word unit with the same pronunciation. 2. The kana-kanji conversion method according to claim 1, wherein when a conversion candidate becomes a first candidate, a conversion candidate with a different pronunciation from the first candidate is selected as the first candidate to be displayed. (3) The non-conversion correction registration means stores the "yomi" on a word-by-word basis, and when a word-based conversion candidate with the same "yomi" appears after storage, it displays the notation of the conversion candidate in kana notation. A kana-kanji conversion method according to claim 1, characterized in that: (4) In the homophone selection correction registration means, “
When the "pronunciation" and "notation" are stored, and a conversion candidate for each word with the same "pronunciation" appears after storage, if the "notation" stored in the registration means can be a conversion candidate of the system, The kana-kanji conversion method according to claim 1, wherein the notation of the conversion candidate is replaced with the "notation" stored in the registration means and displayed. (5) The non-conversion correction registration means and the homophone selection correction registration means are integrally configured, and have the same "yomi" as the "yomi" stored in the homophone selection correction registration means. When a word-by-word conversion candidate appears, if the "notation" stored in the registration means can be a conversion candidate of the system, replace the notation of the conversion candidate with the "notation" stored in the registration means. The kana-kanji conversion method according to claim 1, wherein the kana-kanji conversion method is characterized in that the kana-kanji conversion method is displayed as follows.