JPH0573540A - Kana-Kanji converter - Google Patents

Abstract

(57) [Abstract] [Purpose] In a kana-kanji conversion device that uses a rule dictionary, it is possible to reduce the labor of creating the rule dictionary and the memory capacity of the rule dictionary to improve the conversion efficiency. [Structure] A kana-reading character string is input from an input device, and a kana-kanji conversion program refers to a basic dictionary storing a notation for reading a word, a part-of-speech information section, and a connection table,
Converts the input kana-reading character string into kana-kanji. The rule search program refers to a rule dictionary that stores a pattern of a word string of the basic dictionary and rules having rewriting information thereof, and a part-of-speech information section that collects a part-of-speech ID corresponding to the part-of-speech information, and refers to the rules in the rule dictionary. Search for a match. The rule rewriting program rewrites the content of the corresponding kana-kanji conversion result based on the rule dictionary when a matching rule is retrieved.

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 processing device such as a Japanese word processor.

[0002]

2. Description of the Related Art Conventionally, as a kana-kanji conversion device of this kind, as disclosed in Japanese Patent Laid-Open No. 3-142658, a kana and a corresponding kanji character are basically stored as a pair. It is known to have a dictionary and a rule dictionary in which a notation method in a plurality of specific word strings is stored for each case.

In such a kana-kanji conversion device, first,
Normal kana-kanji conversion is executed with reference to the basic dictionary, and the result of the kana-kanji conversion is stored in the conversion result storage section.
Then, if the stored kana-kanji conversion result matches the pattern in the rule dictionary, the kana-kanji conversion result is rewritten according to the rewriting information of the rule dictionary and displayed on the display device. Moreover, such a series of actions was called rule conversion.

For example, when "Aka-chan ga tatsu" is input, first, the normal kana-kanji conversion with reference to the basic dictionary is performed, and "babies build" is converted. Next, he searched the rule dictionary and, if there was a corresponding pattern, rewrites it as "baby stands" according to the rewriting information.

[0005]

However, in the above method, for example, for a mistranslation result of "baby + adverb + build" such as "baby slowly builds" or "baby firmly builds", The rules for rewriting the wrongly converted "standing" into the correct notation "standing" were created in separate patterns for each adverb and stored in the rule dictionary. Therefore, it is necessary to memorize many similar patterns as the adverbs are different.
The amount of work for creating the rule dictionary has increased, and the memory capacity of the rule dictionary has also increased.

The present invention has been made in order to solve the above-mentioned problems, and stores part-of-speech information representing an arbitrary part-of-speech in a rule dictionary so that one rule corresponds to several patterns having the same sentence structure. Thus, an object of the present invention is to provide a kana-kanji conversion device capable of reducing the capacity of the rule dictionary and performing accurate conversion.

[0007]

In order to solve this problem, a kana-kanji conversion device of the present invention, as shown in FIG. 1, has an input means for inputting a kana-reading character string, a notation for reading a word, etc. And a kana-kanji conversion means for performing kana-kanji conversion by referring to the basic dictionary, a conversion result storage means for storing kana-kanji conversion results by the kana-kanji conversion means, and an output means for outputting the kana-kanji conversion result. And a rule dictionary storing information such as a part of speech of a specific word in the basic dictionary, a pattern of a word string including the word, and rewriting information for the word string, and a conversion result storage unit. Rule searching means for searching a matching pattern by referring to the part-of-speech information from the rule dictionary for contents,
And a rule rewriting unit that rewrites the contents of the corresponding conversion result storing unit based on the rewriting information of the rule dictionary when the matching pattern is searched by the rule searching unit.

[0008]

In the kana-kanji conversion device of the present invention having the above-described structure, first, the kana-kanji conversion string input from the input means is converted by the kana-kanji conversion means by referring to the basic dictionary. Then, the conversion result storage means stores the result of the kana-kanji conversion. The rule dictionary stores rewriting rules corresponding to a plurality of specific word strings including arbitrary part-of-speech information, and if there is a pattern in the result of kana-kanji conversion by the rule searching means that matches the rule dictionary, the rule The rewriting means rewrites the notation according to the rule and outputs it by the output means.

[0009]

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

First, the configuration of the entire kana-kanji conversion device will be described with reference to FIG.

An input device 10 for inputting a character string for kana-kanji conversion is connected to a CPU (central processing unit) 12 for controlling the entire device. The RAM 20 serving as a storage unit is connected to the CPU 12, and the RAM 20
Is a conversion result storage area 22 for storing the result of kana-kanji conversion, a reading input buffer area 24 for storing the input kana reading character string, and the contents of the conversion result storage area 22 for kana-kanji character string. An output buffer area 26 for storing the stored information and a work area 28 for storing pointer information are provided.

In the conversion result storage area 22, as shown in FIG. 3, the entire conversion result is stored by adding information on a word-by-word basis, and is stored in the data 22a representing the word rank and a basic dictionary 42 described later. Word ID 22b, which is a unique numerical value for identifying each word that is present, and word reading 22
c, word notation 22d, part-of-speech ID 22e, which is a unique symbol for identifying each part-of-speech information stored in the part-of-speech information section 48 described later, homonym start ID 22f, and homonym end ID 22g, as data It is remembered for each.

The same-sound word head ID 22f corresponds to the input device 1
This is the same as the character string (word) input from 0, and indicates the smallest numerical value of the word ID 22b stored in the basic dictionary 42. Also, the same phoneme ending ID 22g
Indicates a character having the same sound as a character string (word) input from the input device 10 and having the largest numerical value of the word ID 22b stored in the basic dictionary 42.

ROM 30 for storing programs, dictionaries, etc.
Is connected to the CPU 12, and the program section 32 and the dictionary section 4
It consists of 0 and. The program section 32 stores a kana-kanji conversion program 34, a rule retrieval program 36, and a rule rewriting program 38.

The dictionary unit 40 is composed of a basic dictionary 42, a connection table 44, a rule dictionary 46, and a part-of-speech information unit 48. As shown in FIG. 4, the basic dictionary 42 has a word ID 22b that is a unique identification number of the word.
And reading 22c of the word and notation 22d of the word
And the part-of-speech ID 22e are stored.

The connection table 44 stores data that defines the connection relationship between words by means of part-of-speech information.

As shown in FIG. 5, the rule dictionary 46 stores a plurality of rules, and the content of one rule is composed of a plurality of items 46a to 46c. Generally, the content of one item of the rule is roughly divided into three, and is written in the form of <search factor-search information: rewrite information>, and one rule is created by combining these items. .. The search factor indicates the type of rule search method, and there are four types, U factor, D factor, Y factor, and H factor.

The U factor means a perfect match with the reading and notation of the word stored in the conversion result storage area 22, and the D factor is the reading or the word of the word stored in the conversion result storage area 22. Means the same tone. Further, the Y factor means a match with the reading of the word string (may be one word) stored in the conversion result storage area 22, and the H factor is the part of speech stored in the conversion result storage area 22. It means a match with the ID. Search information 4
6a to 46c represent the corresponding words of each item of the rule, and the specific reading and notation of the word or the part of speech information is entered here.

The rewriting information is either the unchanged marker N, the deleted marker D, or the word string. When the unchanged marker N is stored, the rewriting information matches the item of the rule dictionary 46 in the conversion result storage area 22. Indicates that the word is left as it is and is not rewritten. When the delete marker D is stored, it means to delete the item,
When it is a word string, it means that the word in the conversion result storage area 22 that matches the item of the rule dictionary 46 is rewritten to the word string. The word string referred to here is one or more words stored in the rule dictionary 46 by word ID, and by referring to the basic dictionary 42 based on this word ID, the words in the conversion result storage area 22 are It is possible to set the reading 22c, the same-sound word start ID 22f, the same-sound word end ID 22g, and the like. In FIG. 5, the word I is used for easy understanding.
The part of D is read or expressed not by a numerical value.

In the part-of-speech information section 48, as shown in FIG.
Part-of-speech information 48b referred to via the part-of-speech ID 42a in the basic dictionary 42 is stored.

The output device 50 as an output device for displaying the contents of the output buffer area 26 is the CPU 1
Connected to 2.

Next, the operation of the kana-kanji conversion device configured as described above will be described with reference to the flowchart of FIG.

For example, the reading character string "not clear"
Is input from the input device 10, the kana character code is stored in the reading input buffer area 24 of the RAM 20 (S10). After that, the kana-kanji conversion program 34 of the ROM 30 refers to the basic dictionary 42, the part-of-speech information section 48, and the connection table 44 to refer to the reading input buffer area 24.
The kana character code stored in is converted into a kanji / kana mixed sentence and stored in the reading input buffer area 24 as the kana character code. For example, the kanji / kana mixed sentence “clearly in the city” is converted and stored in the conversion result storage area 22 (S12). Then, the rule search program 36 and the rule rewrite program 38 of the ROM 30 refer to the contents of the conversion result storage area 22 and the rule dictionary 46 to perform the rule conversion process (S14).

Next, a specific process of the rule conversion process will be described with reference to the flow charts of FIGS. 8, 9 and 10.

First, the pointer P1 is set to the first word "clear" in the "clear city" stored in the conversion result storage area 22, and the pointer information of the pointer P1 is RA.
It is stored in the work area 28 of M20 (S30).

Next, it is searched whether or not the rule matching the word pointed by the pointer P1 exists in the rule dictionary 46 (S3).
2). In the rule search subroutine shown in FIG.
First, the pointer P5 is set to the first rule 6- in the rule dictionary 46.
1 is set and the pointer information is stored in the work area 28 (S320). Next, the pointer P6 is changed to the pointer P5.
Is set in the first item 46a of the rule 6-1 pointed to by and the pointer information is stored in the work area 28 (S322).
Next, the pointer P7 is set to the leading word of the kana-kanji conversion result in the conversion result storage area 22 pointed to by the pointer P1 (S324).

Then, it is determined whether or not the search information of the item pointed by the pointer P6 is part-of-speech information (S326). Here, the item 46a pointed to by the pointer P6 is the part of speech information <
H- "to" is an adverb that follows: N> (S326 ・ YE
S), the part-of-speech information part 48 shown in FIG. 6 is sequentially searched by the rule search program 36 from the beginning, and there is an adverb 48c following the matching part-of-speech information "to". Therefore, the pointer P8 is set to the part-of-speech ID (H06). .. Next, the part-of-speech ID indicated by the pointer P8 and the word part-of-speech ID in the conversion result storage unit 22 indicated by the pointer P7 are compared (340). In this case, since the part-of-speech IDs match (S340 / YES), S3
Proceed to 30.

Next, it is determined whether or not the pointer P6 indicates the last item (S330). Since the pointer P6 is not the last item (S330.NO), the pointer P7 is stored in the conversion result storage area 22. Move to the next word "city" after "clear". The pointer P6 is also the next item <Y- in rule 6-1: and,> (46
Then, the pointer information is stored in the work area 28 (S332), and the process returns to the step 326. Here, the item indicated by the pointer P6 <Y-:: ,, >> (46b)
Is not the part-of-speech information (S326 / NO), and the pointer P
Since the item <Y-::, >> (46b) indicated by 6 and the word “city” in the conversion result storage area 22 indicated by the pointer P7 match (YES in S328), the process proceeds to S330.
The item indicated by the pointer P6 is not the last item (S3
30 · NO), the pointer P7 is moved to the word “in” next to “city” stored in the conversion result storage area 22,
The pointer P6 also has the next item <Y-not in rule 6-1:
If not, move to (46c), store the pointer information in the work area 28 (S332), and return to 326.

Thereafter, the same procedure is performed (S32).
6-S332). When the pointer R6 is the last item (YES in S330), the rule flag is turned on, the information is stored in the work area 28 (S342), and the process returns to S32 shown in FIG.

When a matching rule is not retrieved (S
328.NO) moves the pointer P5 to the next rule in sequence until the rule pointed to by the pointer P5 becomes the last rule in the rule dictionary (S344.NO) (S346), and searches for a matching rule (S346). S322-S346). Pointer P
When the rule indicated by 5 is the last rule in the rule dictionary (YES in S344), the rule flag OFF indicating that there is no matching rule is stored in the work area 28 (S348).

Here, since the rule flag of the work area 28 is ON (S34, YES), the rule rewriting process by the rule rewriting program 38 starts. First, the pointer P2 is set to the word "clear" pointed by the pointer P1, and the pointer information of the pointer P2 is stored in the work area 28 (S36). Next, the pointer P3 is set to "in" at the end of the word string "clearly in the city" stored in the conversion result storage area 22 pointed to by the pointer P7, and the information of the pointer P3 is stored in the work area 28 (S3).
8). Next, the pointer P4 is set to the head item of the matched rule 6-1 (46a) in the rule dictionary 42, and the pointer P4 is set.
Information of the pointer is stored in the work area 28 (S4
0). Pointer P in rule 6-1 in rule dictionary 46
4 <H-Adverb following "to": N> (46
Since the rewriting information of a) is attached with the unchanged marker N (S44, YES), the word "clear" is left as it is without rewriting, and the process proceeds to S46.

When the deletion marker D is set at the location of the rewriting information in the item 46b pointed to by the pointer P4 (S44, NO, S48, YES), the word pointed by the pointer P2 in the conversion result storage area 22 is deleted. (S50), S
Proceed to 46. Furthermore, when there is a word string at the location of the rewriting information in the item 46b pointed by the pointer P4 (S44.N
O, S48, NO, S52, YES), the word pointed to by the pointer P2 in the conversion result storage area 22 is rewritten with the word string (S54), and the process proceeds to S46.

In S46, the word pointed by the pointer P2 is compared with the word pointed by the pointer P3. That is, when the word pointed by the pointer P2 and the word pointed by the pointer P3 are equal, it means that the pointer P2 has advanced to the last word. Here, since "clearly" pointed by the pointer P2 is not the last word of the conversion result stored in the conversion result storage area 22 (S46, NO), the pointer P2 is moved to the next word "city", and the pointer is moved. P4 is moved to the next item in the rule dictionary 46 <Set as Y-: Set as> (S56), and the process returns to S44.

Since there is a word string in the item <Y-::, shi> indicated by the pointer P4 (S44, NO, S4).
8. NO, S52, YES), "city" is rewritten as "to" (S54), and the process proceeds to S46. In this way, the contents of the sequential conversion result storage area 22 are processed according to the rule rewriting instruction (S44 to S56).

When the words pointed to by the pointers P2 and P3 become the same, that is, when the word pointed by the pointer P2 becomes the last word stored in the conversion result storage area 22 (S46, YES), If the pointer P1 is not at the end word of the word string stored in the conversion result storage area 22 (S58, NO), the pointer P1 is moved to the next word one by one (S60), and the same rule search process is performed again. Is performed by the rule search program 36. in this case,
The pointer P1 is moved to the second word "city" of the character string "clearly in the city" stored in the conversion result storage area 22 (S58, NO, S60). And S32 to S60
The process of is repeated. Finally, when the pointer P1 is the last word in the conversion result storage area 22 (YES in S58), the rule conversion process ends.

This result is stored again in the conversion result storage area 22, and the contents of the conversion result storage area 22 are stored in the output buffer area 26 and displayed on the output device 50 (S16 in FIG. 7). After that, if the confirmation key is input by the user, the process ends (S18, YES). When a key other than the enter key is input (S18, NO), the process proceeds to the candidate changing process (S20). The result is stored again in the conversion result storage area 22. And the conversion result 22
Of the output device 5 is stored in the output buffer area 26.
0 is displayed (S16). After that, if the confirmation key is input by the user (S18, YES), the process ends.

Conventionally, rules for word strings such as "not clear" and "not jerk" do not say "adverb +" do not follow adverb "" must be made for each adverb. By using as a search factor, it is possible to reduce the labor for creating the rule dictionary and the memory occupied by the rule dictionary.
It is also possible to efficiently perform rule conversion.

Similarly, as in the example of rule 6-2 in FIG.
"It was beautiful when I was young," but "It was beautiful except."
If the conversion result of "other than + adjective verb" such as "Iigaitokantohito" becoming "other than forgiving person" is rewritten as "unexpected and + adjective verb" in this rule conversion Therefore, again, it is not necessary to make a rule to rewrite the same adjectives one by one, and the same effect as described above can be obtained.

Also, in the example of "Aka-chan is slowly standing" and "Aka-chan is firmly standing" in Rule 6-3 of FIG. 5, "Baby (adverb) stands (cut, passed,
It is not necessary to make a similar rule just by replacing adverbs such as "slow" and "firm" when the usual kana-kanji conversion result is made, and the same effect as above can be obtained. it can.

The present invention can be applied not only to the rule conversion using the part-of-speech information as shown in the above example, but also to the rule conversion using the information that can be referred to from the basic dictionary.

The number of rules and the number of items in the rule dictionary are not limited to those in this embodiment.

[0042]

As is clear from the above description,
In the rule conversion process, the kana-kanji conversion device of the present invention had to make many similar rules by simply replacing words of the same part of speech, but by using part-of-speech information as search information, Since it is now possible to make rules together, it is possible to reduce the labor for creating the rule dictionary and also reduce the memory capacity of the rule dictionary. Also, one rule can be applied to many sentence examples.

[Brief description of drawings]

FIG. 1 is a configuration diagram of the present invention.

FIG. 2 is a block diagram of a control unit of the kana-kanji conversion device according to the present embodiment.

FIG. 3 is a conceptual diagram of a conversion result storage area according to the present embodiment.

FIG. 4 is a conceptual diagram of contents of a basic dictionary of this embodiment.

FIG. 5 is a conceptual diagram of contents of a rule dictionary of this embodiment.

FIG. 6 is a conceptual diagram of contents of a part-of-speech information unit of the present embodiment.

FIG. 7 is a flowchart of the kana-kanji conversion device of the present embodiment.

FIG. 8 is a flowchart of rule conversion of this embodiment.

FIG. 9 is a flowchart of rule conversion of this embodiment.

FIG. 10 is a flowchart of a rule search according to this embodiment.

[Explanation of symbols]

 10 Input Device 12 CPU 20 RAM 22 Conversion Result Storage Area 24 Reading Input Buffer Area 26 Output Buffer Area 30 ROM 31 Program Section 34 Kana-Kanji Conversion Program 36 Rule Search Program 38 Rule Rewriting Program 40 Dictionary Section 42 Basic Dictionary 44 Connection Table 46 Rule Dictionary 48 part-of-speech information section 50 output device

Claims (1)

[Claims] 1. A kana-kanji conversion unit for inputting a kana-yomi character string, a basic dictionary storing notations for reading the words, kana-kanji conversion means for converting kana-kanji by referring to the basic dictionary, and the kana-kanji conversion. A kana-kanji conversion device comprising conversion result storage means for storing conversion results by means, and output means for outputting the kana-kanji conversion result, including part-of-speech information of a specific word in the basic dictionary, and the word. A rule dictionary that stores a pattern of a word string and rewriting information for the word string, and a rule search for searching for a matching pattern in the rule dictionary for the contents of the conversion result storage means by referring to part-of-speech information Means, and when the matching pattern is searched by the rule search means, the contents of the corresponding conversion result storage means are converted into rewriting information of the rule dictionary. Kana-kanji conversion apparatus being characterized in that a rule rewriting means for rewriting Zui.