JPH10124498A - Japanese syllabary/chinese character converting device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a device in which conversion precision is improved by increasing the number of co-occurrence data used at the time of conversion by referring to co-occurrence data by plural methods concerning a word for which the co-occurrence data can not be obtained at the time of initial reference. SOLUTION: A Japanese syllabary/Chinese character conversion controlling part 8 designates a word for which co-occurrence data should be refereed to for a co-occurrence data referring part 4. The co-occurrence data referring part 4 refers to data related with the designated word from a co-occurrence data dictionary 7, changes a method for collating data when the objective data can not be obtained, and refers to the co-occurrence data again. Thus, it is possible to refer to the co-occurrence data which can not be obtailed at the time of initial reference. For example, 'SANNIN FUKUSHIMAHO SEN (Japanese syllabary)' is converted into 'SANIN FUKUSHIMA HOSEN (Chinese character)' in a conventional Japanese syllabary/Chinese character converter using a first candidate, and it is converted into 'SANIN FUKUSHIMA HOSEN (Shinese character)' in this device by using co-occurrence data, 'SANIN' and 'HOSEN'.

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 device for converting input kana data into kanji data.

[0002]

2. Description of the Related Art Conventionally, in a kana-kanji conversion device, when changing input kana data to kanji data, a certain data is selected and displayed from a plurality of conversion candidates. The conversion result is corrected by selecting the correct kanji data from the conversion candidates.

[0003] Improving the correct answer rate (conversion accuracy) of the first candidate is one of the important issues in the kana-kanji conversion apparatus. For that purpose, co-occurrence data representing the co-occurrence relation of a plurality of words is used. Conversion has also been performed.

[0004]

In the conversion using the co-occurrence data, the conversion accuracy depends on the number of the corresponding co-occurrence data. If the amount of the co-occurrence data to be prepared is large, the conversion accuracy is increased accordingly. improves.

However, the capacity of a storage means such as a memory is limited, and a continuous expression of a plurality of nouns and a co-occurrence relationship of various linguistic phenomena, which are innumerably present in Japanese, are required.
It is impossible to register all as co-occurrence data.

Accordingly, it is an object of the present invention to provide a kana-kanji conversion apparatus in which the number of co-occurrence data used at the time of conversion is increased by devising a reference procedure for co-occurrence data and conversion accuracy is improved.

[0007]

A kana-kanji conversion apparatus according to the present invention comprises a conversion dictionary storing reading of Japanese words, kanji notation, part of speech information and semantic classification information, and a dictionary for searching a conversion dictionary for a specified word. When the search unit and the co-occurrence data dictionary that stores the co-occurrence data indicating the co-occurrence relationship between words and the specified words are collated, if the co-occurrence data is not obtained, A co-occurrence data reference unit that changes the data collation method again for collation based on the control of the kana-kanji conversion control unit, a dictionary search unit, a co-occurrence data reference unit, and a kana-kanji conversion control unit that controls the data flow And

[0008] With this, for a word for which co-occurrence data could not be obtained by the initial reference, more co-occurrence data can be referred to by referring to the co-occurrence data by a plurality of methods, and conversion accuracy can be improved. Can be enhanced.

[0009]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention according to claim 1 of the present invention provides an input means for inputting a kana character string and selecting a conversion candidate, an input kana character string, a conversion candidate, and a converted kana-kanji character. A conversion data storage unit that stores a character string, a conversion dictionary that stores reading of Japanese words, kanji notation, part of speech information, and semantic classification information; a dictionary search unit that searches a conversion dictionary for a specified word; A co-occurrence data dictionary that stores co-occurrence data indicating the co-occurrence relationship of words, a co-occurrence data reference unit that performs collation of co-occurrence data for a specified word, an input character string, conversion candidates, and conversion results Display means for displaying, and a kana-kanji conversion control section for controlling the flow of data stored in the dictionary search section, the co-occurrence data reference section, and the conversion data storage section, and in the co-occurrence data reference section, Co-occurrence data If you did, by changing the method of collating data, it was decided to carry out a reference again co-occurrence data.

Therefore, the kana-kanji conversion control unit specifies a word that refers to the co-occurrence data to the co-occurrence data reference unit, and the co-occurrence data reference unit uses the co-occurrence data dictionary to store data relating to the specified word. If the reference data is not obtained, the collation method of the data is changed, and the co-occurrence data that cannot be obtained by the initial reference can be referred to by referring to the co-occurrence data again. .

[0011] According to a second aspect of the present invention, when the co-occurrence data reference unit does not obtain co-occurrence data for a series of noun words consisting of a plurality of nouns, the nouns forming the noun series We decided to refer to the co-occurrence data again for some combinations of.

Therefore, in the co-occurrence data reference section,
If co-occurrence data for a series of nouns consisting of multiple nouns is not available, referencing the co-occurrence data for some combinations of the nouns that make up the noun continuation allows for initial reference. The co-occurrence data relating to the missing noun sequence can be referred to.

According to a third aspect of the present invention, when co-occurrence data of a specified word is not obtained in the co-occurrence data reference section, the co-occurrence data is referred to and the semantic classification is performed again. It refers to the co-occurrence data of the same word.
Therefore, when co-occurrence data is not obtained in the co-occurrence data reference unit, by referring to the semantic classification information and referring to the co-occurrence data again, the co-occurrence data for the related words having the same semantic classification can be obtained. Can be referenced.

According to a fourth aspect of the present invention, when co-occurrence data for an input of "A to B" type is not obtained in the co-occurrence data reference section, "A" and "B"
The co-occurrence data is referred to again for the word "AB" which is a sequence of "." Therefore, if the co-occurrence data reference unit does not obtain the co-occurrence data for the input of “A to B” type, the co-occurrence data is re-targeted to the word “AB” that is a series of “A” and “B”. , It is possible to refer to the co-occurrence data related to the word representing the operation, which could not be referred to by the initial reference.

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a functional block diagram of a kana-kanji conversion device according to an embodiment of the present invention.

In FIG. 1, reference numeral 1 denotes an input means for inputting a kana character string and an instruction for selecting a conversion candidate; and 2, a conversion means for storing an input kana character string, a conversion candidate, and a converted kana kanji character string. A data storage unit 3 is a dictionary search unit that searches the conversion dictionary 6 for a specified word. 4 is a co-occurrence data reference unit that performs collation of co-occurrence data on a specified word. , Conversion candidates, and conversion means for displaying conversion results, etc., 6 is a conversion dictionary storing reading of Japanese words, kanji notation, part of speech information, semantic classification information, etc., and 7 shows a co-occurrence relationship between words. A co-occurrence data dictionary storing the co-occurrence data, a kana-kanji conversion control for controlling the operation of each function of the dictionary search section 3 and the co-occurrence data reference section 4 and the flow of data stored in the conversion data storage section 2 Department.

In the circuit block diagram shown in FIG. 2, 9 is a keyboard, 10 is a CPU (Central Processing Unit), and 11 is C
RT (Cathode Ray Tube Display), 12 is a ROM (Read Only Memory), and 13 is a RAM (Random Access Memory).

The input means 1 shown in FIG. 1 is operated by the keyboard 9, the converted data storage unit 2 is operated by the RAM 13, the conversion dictionary 6, the co-occurrence data dictionary 7 is operated by the ROM 12, the dictionary search unit 3, the co-occurrence data reference unit 4, The kana-kanji conversion control unit 8
When the CPU 10 executes the operation program stored in the ROM 12 while exchanging data with the ROM 12 and the RAM 13, the display unit 5 displays the CRT 11.
, Respectively.

The operation of kana-kanji conversion in the kana-kanji conversion apparatus of the present embodiment configured as described above will be described below.

The kana character string input by the input unit 1 is stored in the conversion data storage unit 2 by the kana-kanji conversion control unit 8 in accordance with a user's conversion instruction.

The kana-kanji conversion control unit 8 performs a dictionary search on the conversion dictionary 6 by the dictionary search unit 3 on a partial character string in the kana character string stored in the conversion data storage unit 2 and obtains a search result. Is stored in the conversion data storage unit 2.

Next, the kana-kanji conversion control unit 8 determines the division of the words of the input kana character string based on the search result by the dictionary search unit 3. When a plurality of homophone words are conversion candidates in the search result of the dictionary search unit 3 in each of the divided words, the kana-kanji conversion control unit 8 controls the co-occurrence data reference unit 4 to store the co-occurrence data dictionary 7 Refer to and, based on the relevant data,
The first conversion candidate is determined.

The kana-kanji conversion control section 8 displays the determined kana-kanji character string on the display means 5, and displays, changes and confirms conversion candidates in accordance with a user's instruction.

The operation of referring to co-occurrence data performed by the co-occurrence data reference section 4 under the control of the kana-kanji conversion control section 8 will be described below with reference to the flowchart of FIG.

First, in step S 1, the kana-kanji conversion control unit 8 passes a kana character string of a word to be referred to in the conversion target sentence stored in the conversion data storage unit 2 to the co-occurrence data reference unit 4. In step S2, the co-occurrence data reference unit 4
Selects data beginning with the specified reading from the co-occurrence data stored in the co-occurrence data dictionary 7.

In step S3, the co-occurrence data reference unit 4
In step S1, information as to whether or not the specified word is a definite word is confirmed. If the word is a definite word, the process proceeds to step S4. In step S4, in the co-occurrence data selected in step S2, a word whose kanji notation of the first word specified by reading is different from a fixed word already determined as a conversion candidate of the word is deleted, and the process proceeds to step S5. . If it is determined in step S3 that the designated word is not a definite word, the process directly proceeds to step S5. In step S5, the kana-kanji conversion control unit 8 stores the co-occurrence data whose head words match each other as candidate co-occurrence data in the conversion data storage unit 2.

In step S 6, for each of the co-occurrence data stored in the conversion data storage unit 2, the character string following the specified word in reading the co-occurrence data and the conversion target character string stored in the conversion data storage unit 2 Compares the character string following the specified word. In step S7, it is confirmed whether there is data in which the two kana character strings match in the comparison in step S6. If there is no matching data, in step S9, the data collation method is changed again, and the data is referenced.
If there is matching data, the co-occurrence data is stored as the corresponding data in the converted data storage unit 2 in step S8, and the process ends.

The operation of the three embodiments of data reference performed in step S9 will be described below with reference to the flowcharts of FIGS. 4, 5, and 6, respectively.

(Embodiment 1) First, Step 9A-1
It is determined whether or not the designated word is a word that can be a noun. If the specified word cannot be a noun, the processing ends because this data reference method is not applied. If the specified word can be a noun, in step 9A-2, the number of consecutive nouns that can be formed starting from the specified word (hereinafter, this number is referred to as N) is obtained.

For example, if the input kana character string is "sanin / fukushima / hosen" and "sanin" is a designated word, the specified word "sanin" can be a noun in step 9A-1. , And proceed to Step 9A-2. In addition, since each of “sanin”, “Fukushima”, and “hosen” can be nouns, the number N of consecutive nouns is obtained in step 9A-2.

At step 9A-3, if N <3, the processing is terminated because noun continuations having different constituent elements from the current one cannot be generated. If N ≧ 3, the process proceeds to step 9A-4.

Steps 9A-4 to 9A-12
Generates a noun sequence to be referred to. Step 9A
At -4, a variable s = 1 that defines the number of nouns from the beginning is set. If s ≦ N−2 in step 9A-5, the process proceeds to step 9A-6, in which a variable t representing the number of nouns connected to the first s is set to t = s + 1.

If t> N in step 9A-7, there is no corresponding noun, so the process proceeds to step 9A-13.
In step 9A-13, the number s of nouns from the beginning is set to 1
And returns to step 9A-5. Step 9A-7
If t ≦ N, the process proceeds to step 9A-8, where a series of nouns is generated by connecting the first s nouns and the t-th noun from the top.

For example, in the above example of "sanin / fukushima / hosen", when s = 1 and t = s + 1 = 2,
Generate "Sanfu Fukushima".

In step 9A-9, the words generated in step 9A-8 are collated with the candidate co-occurrence data stored in the conversion data storage unit 2 in step S5. In step 9A-10, the presence or absence of the corresponding data in the collation of the co-occurrence data in step 9A-9 is checked. If there is no such data, the process proceeds to step 9A-12. If there is matching data, the co-occurrence data is stored in the converted data storage unit 2 as corresponding data in step 9A-11.

In step 9A-12, t indicating the number of the next noun from the beginning is incremented by one, and the process returns to step 9A-7.

For example, in the above example of "sanin / fukushima / hosen", in step 9A-7, t (= 3)
Since ≦ N (= 3), the process proceeds to step 9A-8, where s =
1. From “t = 3”, “saninhosen” is generated.

If s> N-2 in step 9A-5, generation of the noun sequence to be referred to has been completed.
The process ends.

For example, "sanin / fukushima / hosen"
In the example of (1), N = 3, and “samfukushima” is generated at s = 1 and t = 2, and “small book” is generated at s = 1 and t = 3. In the case of s = 2, the same word as the first word "saninfukushimahosen" will be generated. Therefore, s (= 2)> N−2 (= 1) in step 9A-5.
If, the process ends.

Using this co-occurrence data collation step, in step 9A-9, if collation of "saninhosen" is successful, co-occurrence data "saninhosen (secondary election of the Upper House)" is obtained. As a result, in the conventional Kana-Kanji conversion device using the first candidate in "Sanfu Fukushima Hosen", it was converted to "San-in Fukushima Yasen" etc.
By using co-occurrence data for “sanin” and “hosen”, the kana-kanji conversion device of the present embodiment can convert the data to “Sanin Fukushima Supplementary Selection”.

(Embodiment 2) Next, a second method for referring to co-occurrence data will be described with reference to FIG.

First, in step 9B-1, the candidate co-occurrence data stored in the converted data storage unit 2 in step S5 is extracted. In Step 9B-2, Step 9B-
One candidate co-occurrence data extracted in step 1 is obtained. Step 9
In B-3, the semantic classification information of the word next to the designated word in the candidate co-occurrence data obtained in step 9B-2 is obtained. In step 9B-4, the semantic classification information of the conversion candidate of the word next to the designated word stored in the conversion data storage unit 2 is obtained.

In step 9B-5, the semantic classification information of the co-occurrence data obtained in step 9B-3 is compared with the semantic classification information of the conversion candidate obtained in step 9B-4. Proceed to step 9B-6. Step 9B
In -6, in the co-occurrence data to be compared in step 9B-5, the data in which the word referring to the semantic classification information is changed to a conversion candidate whose semantic classification information matches is converted as the corresponding co-occurrence data in the conversion data storage unit. 2 is stored.

If the semantic classification information is not equal in step 9B-5, the process returns to step 9B-2 to obtain the next candidate co-occurrence data. If the next candidate co-occurrence data is not obtained in step 9B-2, the collation for all the data has been completed, and the process ends.

For example, if the input character string is "Kyoka / Kikan" and "Kyoka" is a designated word, step 9B
-1, the candidate co-occurrence data stored in the conversion data storage unit 2, "Kyoka / renshu (enhancement / practice): 5/1
3 ”,“ Today / Kaidai (enhancement / enlargement) 5/2
2 ”,“ Kyoka / Gekkan (enhanced / monthly) 5/1
8 ", ... are taken out.

In step 9B-2, the first candidate co-occurrence data "Kyouka / renshu (enhancement / practice): 5/13"
Get. If the semantic classification information is added after the kanji notation of the co-occurrence data, in step 9B-3, the semantic classification information 1 of the word “renshu” following the designated word “kyoka”
Get 3.

In step 9B-4, the conversion data storage unit 2
, The conversion candidate of the next word “Kikan” after the designated word, the institution (10), the period (18), the quarterly (34)... In step 9B-5, the semantic classification information (13) of the co-occurrence data and the semantic classification information (10, 1,
8, 34).

When the same processing is performed for all the candidate co-occurrence data, the reference to the candidate co-occurrence data “Kyoka / Gekikan (enhanced / monthly) 5/18” refers to Step 9B-
In 5, the semantic classification information (18) of the co-occurrence data matches the semantic classification information (18) of the conversion candidate “period”.

When co-occurrence data of “Kyoka” and “Gekkan” exist, “Month” and “Period” have the same semantic classification information. Period) 5/18 "is obtained. As a result, in the input kana character string "Kyokakikan", the conventional kana-kanji conversion using the first candidate was converted to "Strengthening institution", etc., but co-occurred in "Kyouka" and "Kikan" Using the data, the kana-kanji conversion apparatus according to the present embodiment can convert the data into the “enhancement period”.

(Embodiment 3) Next, a third method for referring to co-occurrence data will be described with reference to FIG.

First, in step 9C-1, it is determined whether the character string from the designated word in the conversion target sentence stored in the conversion data storage unit 2 is "A (noun) to B (noun)". If not, the data reference method is not applied, and the process ends. When it is determined that this data reference method is applied, the process proceeds to step 9C-2. In step 9C-2, “A” stored in the conversion data storage
A character string “AB” is generated from an input character string of the “do B” type.

In step 9C-3, the word generated in step 9C-2 is collated with the candidate co-occurrence data stored in the conversion data storage unit 2 in step S5. In step 9C-4, the presence or absence of the corresponding data in the collation of the co-occurrence data in step 9C-3 is checked, and if there is such data, in step 9C-5, the co-occurrence data is used as the corresponding data in the conversion data storage unit. 2 and the process ends.

For example, the input kana character string "Choose Seikaku" is of the type "A (Sekaiku) to B (choice)", so that step 9C-1 and step 9C-2
Proceed to. In step 9C-2, “Squareness” is generated from “Squareness”.

In step 9C-3, the word "seikakuchosa" generated in step 9C-2 is collated with the candidate co-occurrence data stored in the converted data storage unit 2 in step S5. If the co-occurrence data “character search” is collated in step 9C-3, the co-occurrence data is stored as the corresponding data in the converted data storage unit 2 in step 9C-5, and the process ends.

As a result, in the input kana character string "Sho-ku-se-ku-sakuru", the conventional kana-kanji conversion device using the first candidate converts the character string to "Check for accuracy" or the like. By using the co-occurrence data for “chosa”, the kana-kanji conversion device of the present embodiment can be converted to “investigate personality”.

In the present embodiment, a plurality of data reference methods have been described separately in different embodiments. However, in the embodiment, a plurality of data reference methods may be used simultaneously. No problem.

[0057]

According to the present invention, data that has not been referred to in the first co-occurrence data reference can be obtained as co-occurrence data. Can be improved.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a kana-kanji conversion device according to an embodiment of the present invention.

FIG. 2 is a circuit block diagram of a kana-kanji conversion device according to an embodiment of the present invention.

FIG. 3 is a flowchart illustrating an operation of referring to co-occurrence data according to an embodiment of the present invention;

FIG. 4 is a flowchart showing a data reference operation according to the first embodiment of the present invention;

FIG. 5 is a flowchart showing a data reference operation according to the second embodiment of the present invention;

FIG. 6 is a flowchart illustrating a data reference operation according to the third embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Input means 2 Conversion data storage part 3 Dictionary search part 4 Co-occurrence data reference part 5 Display means 6 Conversion dictionary 7 Co-occurrence data dictionary 8 Kana-kanji conversion control part

Claims (4)

[Claims] An input means for inputting a kana character string and selecting a conversion candidate, a conversion data storage unit for storing an input kana character string, a conversion candidate, and a converted kana-kanji character string, and a Japanese word , A conversion dictionary that stores kanji notation, part-of-speech information, and semantic classification information, a dictionary search unit that searches the conversion dictionary for a specified word, and co-occurrence data indicating the co-occurrence relationship between words. A co-occurrence data dictionary, a co-occurrence data reference unit for collating co-occurrence data with respect to a specified word, display means for displaying an input character string, a conversion candidate, and a conversion result; And a kana-kanji conversion control unit for controlling the flow of data stored in the conversion data storage unit. In the co-occurrence data reference unit, co-occurrence data of a specified word could not be obtained. If the date Kana-kanji conversion apparatus of collation method to change the, and performs reference again co-occurrence data. 2. If the co-occurrence data reference unit does not obtain co-occurrence data for a series of nouns consisting of a plurality of nouns, the co-occurrence data reference unit re-selects some combinations of nouns forming the noun series. The kana-kanji conversion device according to claim 1, wherein the co-occurrence data is referred to. 3. If the co-occurrence data reference unit does not obtain co-occurrence data of a specified word, the co-occurrence data is referred to, and the co-occurrence data of the word having the same semantic classification is referred to again. The kana-kanji conversion device according to claim 1 or 2, wherein: 4. In the co-occurrence data reference section, "A
2. The co-occurrence data is referred to again for a word "AB" formed by a continuation of "A" and "B" when co-occurrence data for an "type" input is not obtained. 2. The kana-kanji conversion device according to claim 2 or 3.