JPS59103136A - Kana (japanese syllabary)/kanji (chinese character) processor - Google Patents

Abstract

PURPOSE:To perform the KANA/KANJI conversion with high efficiency while reducing the load of an operator by discriminating a homophone and a word of same type with a function that learns the homophone for each paragraph in case the homophone exist. CONSTITUTION:A keyboard 11 feeds a Japanese sentence as a KANA sentence and stores it temporarily in an input character buffer register 12. A word dictionary storage circuit contains an independent word dictionary 18 containing the output order information of homonyms, a prefix list 19, a suffix list 20, etc. A control circuit 13 retrieves a word dictionary circuit and converts the contents of a buffer circuit into a sentence containing KANA and KANJI. Then the information on parts of speed, prefixes and suffixes are stored in buffers 25, 27 and 29 respectively, and at the same time the fixed information is stored opposite to the input character string of the keyboard 11. For these buffer output characters, the part-of-speech code is recorded in place of the output character string for a specific part of speech.

Description

[Detailed Description of the Invention] Technical Field The present invention relates to a kana-kanji conversion processing device used in Japanese word processors, etc., and particularly relates to a kana-kanji conversion processing device used in Japanese word processors, etc. The present invention relates to a kana-kanji conversion processing device that can improve the automatic discrimination rate of homophones and homophones while reducing the burden on the operator by learning homophones in units.

Conventional technology Kana-Kanji conversion processing (hereinafter also simply referred to as "conversion processing")
Various types of devices have been proposed. One such device is the device previously proposed by the present applicant in Japanese Patent Application No. 149737/1983 titled “Kana-Kanji Conversion Processing Device.” This device includes a means for inputting a Japanese sentence as a kana sentence, a means for temporarily storing a character string inputted by the input means, a word dictionary storage means, and a means for searching the word dictionary storage means for temporarily storing the character string. In a kana-kanji conversion processing device having a kana-kanji conversion control circuit for converting the contents of a means into a sentence mixed with kanji, the word dictionary includes output ranking information of homophone and allograph words and stores words extracted from the word dictionary. a word buffer memory; means for outputting the contents of the word buffer memory according to the output order information;
Means for replacing output ranking information between a word selected and output from the output and a homophone of the word output immediately before it, and a means for replacing the output ranking information with the replaced output ranking information at a predetermined location in the word dictionary. It is a kana-kanji conversion processing device that is characterized by having means for writing in words, and can be said to have a learning function on a word (independent word) basis.

However, when trying to obtain a conversion result such as "What is fire?", the following problem arises.

That is, as a result of using an independent word dictionary such as the example shown in Figure 1 for the input of ``Hito wa'', the display order of conversion candidates using the longest match method is ``Hito wa'' → ``-wa''. → "What is fire" → "What is fire", and the next candidate key (update key) must be pressed many times.

To solve this problem, there is a device proposed by the present applicant as Japanese Patent Application No. 57-76076 titled "Kana-Kanji Conversion Processing Device." This device includes a means for inputting Japanese sentences as kana sentences, a means for temporarily storing character strings input by the input means, a word dictionary storage means including output ranking information of homophone words, and A kana-kanji conversion control circuit for searching a dictionary storage means and converting the contents of the temporary storage means into a sentence mixed with kanji, a means for displaying a conversion result by the kana-kanji conversion control circuit, and a means for displaying final information of the conversion result. The kana-kanji conversion processing device ft1J is characterized in that the kana-kanji conversion processing device has means for storing the confirmed information in correspondence with at least an input character string, and has a so-called homophone buffer memory. It can be said that it is a thing.

However, even in this device, words with prefixes or suffixes (these are called "homogeneous words") are not stored in the homophone buffer memory, so the same analysis must be performed each time they are input. There is a problem.

Purpose The present invention was made in view of the above circumstances, and its purpose is to solve the problems with the above-mentioned device and to reduce the burden on the operator by extending the concept of homonyms to the range of the same words. An object of the present invention is to provide a kana-kanji conversion processing device capable of improving the automatic discrimination rate of homophones while reducing the noise.

composition Hereinafter, the configuration of the present invention will be explained in detail using examples.

FIG. 2 is a block diagram of a kana-kanji conversion processing device showing one embodiment of the present invention. The device of this embodiment classifies Japanese words constituting an input sentence into independent words, prefixes, suffixes, and adjunct words, breaks down the input sentence into division units centered on one independent word, and divides the input sentence into kana characters. , the input character string is stored in a temporary storage means, and the stored content is compared with the content of a dictionary that stores each of the independent words, prefixes, suffixes, and attached words. This figure shows an example in which the present invention is applied to a kana-kanji conversion processing device.

In addition, in the description of this embodiment, the parting unit (also referred to as "bunsetsu") refers to the following.

[Prefix] (Independent word) [Suffix] [Adjunct word] 8 Here, there must be one independent word, the words enclosed in [ ] may be omitted, and the attached word (marked with *) ) may have multiple eccentric speeds.

In FIG. 2, 11 is an input keyboard device, 12 is an input character buffer register for temporarily storing input character strings, 13 is a control circuit, 14 is an output editor, 15 is a display device, and 16 is a printer. 17 to 24 indicate dictionary documents, 17 is a connection matrix table, 18 is an independent word dictionary, 19 is a prefix table, 20 is a suffix table, 21 is a word usage table, 22 is an adjunct word table, and 23 is a list of attached words. This is a katakana hirakana correspondence table. The contents of these dictionaries will be described later.

Further, 24 is an output sentence register, 25, 27, and 29 are output buffer memories for independent words, prefixes, and suffixes, respectively, and 26, 28, and 30 are rank controllers.

31 is a homophone phrase buffer memory.

Hereinafter, the specific contents of the second dictionary table will be explained using an example. The contents of the dictionary or table are as follows.

O reading kana: The part that matches the input character string 0 Notation: Characters to be output when a specific word is determined O Part of speech, Conjugation: Record the part of speech and conjugation of each independent word with a number O Output order: At a certain point 0) Independent Word Dictionary 18 As part of the content is shown in Figure 3, among independent words,
Nouns (including numbers and proper nouns),
It memorizes things that belong to the following categories: verbs, adjectives, adjectives, adjectives, adverbs, interjections, etc. Among these, the verb is only the stem, but the first-stage conjugated verb includes the inflected part, and the conjugated verb ``suru'' and the force conjugated verb ``kuru'' include the entire conjugated form. .

Q) Prefix Table 19 As part of its contents is shown in Figure 4, it stores prefixes attached to nouns, etc. of independent words.

(3) Suffix Table 20 As part of its contents is shown in Figure 5, it stores suffixes attached to independent word nouns, etc.

(4) Word Conjugation Table 21 As part of its contents is shown in Figure 6, the conjugation endings are stored for each classification of verbs, adjectives, parts of speech of adjective verbs, and conjugations. The five-stage conjugation endings of verbs are shown in Figure 7.
Figure 8 shows the single-stage conjugation endings, and Figures 9 and 10 show the conjugation endings of adjectives and adjective verbs, respectively.

(51 Adjunct word classification table 22 consists of a major classification table 22A and a minor classification table 22B,
The major classification table 22A is classified into containers of formal nouns, auxiliary verbs (formal verbs), auxiliary verbs, and particles, as part of its contents are shown in Figure 11, and corresponds to the minor classification table 22B by number. It's a shame. As part of the contents of the minor classification table 22B is shown in FIG. 12, the contents are classified by formal noun container, auxiliary verb, auxiliary verb container conjugation, and particle container conjugation.

(6) Connection matrix table 17 This table shows the grammatical connection relationship between independent words and attached words by arranging preceding words in rows and succeeding words in columns. Element C1j in row 1 and column j is It is defined as follows.

←) If Cg=1, the word in the first row can be connected to the word in the jth column (b) If c1j=o, the word in the first row cannot be connected to the word in the jth column. Some of them are shown. Note that the column numbers in FIG. 13 are shown in hexadecimal numbers for every four columns.

Further, the serial numbers in the major classification table 22A of the adjunct words correspond to the column numbers.

FIG. 14 shows a model of the step of selecting adjunct words using connection matrix table 17.

Furthermore, the connection matrix table 17 has an item for termination possibility, and when this is '1'', it indicates that there is a possibility of termination as the segment.

The operation of this embodiment configured as described above will be described below.
explain.

(1) Input operation Input is performed using the keyboard device 11. An example of the character arrangement of the keyboard device 11 is shown in FIG. 15, but the arrangement is not limited to this.

This keyboard device is configured to output phrase break information by pressing the shift key and a special key marked with a circle or a mark. By combining the shift key (hereinafter referred to as "■") and the special key, the following conversion mode can be specified.

(b) Immediately after a phrase that requires conversion into a sentence containing kanji, press the ■ key, and then press the [Ao] key.

(Example) Keystroke order Conversion transfer output I am O → I am Katsukou O Or press your mouth.

(Example) Keystroke sequence Output after conversion Lonely ■ Mouth → Lonely School/Kesu ■ Enclosure → School Bus ■ Conversion processing operation First, "What is fire? Fire is something that humans cannot live without." An example of creating a sentence is shown below. In the following explanation, the above-mentioned 09° ■ diagram, ■ enclosure, etc. will be represented by the bunsetsu delimiter ■.

In addition, the homophone phrase buffer 31, which is a feature of the present invention, can store phrases that have homophones and would be displayed at a lower order in normal conversion, or prefixes and suffixes, as shown in an example in FIG. This is to register phrases that include words that have the same pronunciation and notation. Initially, nothing is registered in this homophone phrase buffer 31, but in the above-mentioned situation, used phrases are registered, including their frequency, if necessary.

Returning to the example sentence for m1 above, the operator uses the keyboard device 11.
Therefore, one is [phase] something. Hitoke [Ao] Something that a person cannot do without [Ao].

When the first phrase "Hitoke" is input and the phrase delimiter information is input, first, the homophone phrase buffer 31 is searched. However, at this stage, nothing is stored in the homophone phrase buffer 31, so the routine moves on to a normal dictionary search.
It is translated as ``人は''.

This is not what the operator intended, so when the operator presses the update key, "-wa" is displayed, and when the operator presses the update key again, "non and what" is displayed, and then "fire and what" is displayed. . This is what the operator intended, so he moves on to input the next clause and says, "Something."
Enter. This input operation (strictly speaking, [something. J's [
With the input operation of ``J''), the previously input ``Hitohako'' and its output display ``Fire and what'' are registered in the homophone phrase buffer 31. Next, upon completion of the input of "something.", after searching the homophone phrase buffer 31 (there is no corresponding item in the homophone phrase buffer 31), it is converted into "something." by a normal dictionary search.

Next, when "Hitoke" is input and phrase delimiter information is input, the homophone phrase buffer 31 is searched for the character string "Hitoke". As a result, the previously stored "What is Fire" can be retrieved. Furthermore, if you input "Hitoto ■", the character string "Hitoto" will be displayed in the homophone phrase buffer 3.
After searching for 1, a normal dictionary search is performed to output and display the words ``for people''.

Next, we will take the case of creating a contract as an example.

A contract generally consists of around 10 or more clauses. In such a case, in conventional processing, for example, for the human power of ``daiichijo■'', it is processed through the process of processing it as an independent word, an independent word + attached word, and an independent language expert suffix ten attached words. , the prefix is processed as an independent word + an adjunct word, "dai" is taken out as a prefix, rlJ is treated as an independent word (numerical word), and "jiyou" is treated as an adjunct word. and adjunct word table), but since there is no connection that can be established, processing begins with the prefix being treated as a prefix + an independent booklet suffix. Here, as the suffix "jiyo", "ba" is first obtained, but this is not the purpose, so by operating the next candidate key, "jo" is obtained.
get. By inputting the next character of "Article 1", "Daichijo" and "Article 1" are registered in the homophone phrase buffer 31.

However, in the same document, the word "Article 1" does not appear again, but instead "Article 2" and "Article 3".

As mentioned above, even if "Article 1" is registered, it is not used in the document, but "Article 2", "Article 3", etc. In order to obtain ・, the same analysis as before may be performed each time. Therefore,
When registering "Daiichi Higashi" in the homophone phrase buffer 31, instead of registering the numerical part as is, the part-of-speech code (
If there are numbers in the input character string, the numbers are converted into part-of-speech codes (number words) and then the homophone phrase buffer 31 is searched. By doing this, no matter what the ) character part is, if the readings of the prefix, suffix, or both match, the contents of the homophone phrase buffer 31 can be retrieved. FIG. 17 is a program flowchart of the above processing. By performing the above-described processing, it is possible to reduce the capacity of the same sound phrase buffer, and the processing speed is improved.

As mentioned above, when homophones occur in a phrase and the first candidate does not give a correct answer, they are registered one after another in the homophone phrase buffer 31, and when this homophone phrase buffer 31 becomes full, , it is a good idea to delete the already registered homophones and register the latest homophones there. How to delete: (1) Delete the oldest item first.

■) Delete phrases that are least frequently used before a certain registration order (for example, half the number of registered phrases).

The following methods are possible, but are not limited to these methods.

In the above embodiment, phrases that would be displayed at a lower order in normal conversion processing are registered in the homophone phrase buffer 31. It is also possible to decide as appropriate, such as below the candidate or below the third candidate.

In addition, in the above example, an example was shown in which words (clauses) that include number words are treated as homogeneous words, but in addition to number words, proper nouns such as people's names and place names may also be treated in the same way. .

An example is shown in FIG.

Effects As described above, according to the present invention, there is provided a means for inputting a Japanese sentence as a kana sentence, a means for temporarily storing a character string inputted by the input means, and a word including output ranking information of homophone and allograph words. a dictionary storage means, a kana-kanji conversion control circuit that searches the word dictionary storage means and converts the contents of the temporary storage means into a sentence mixed with kanji, a means for displaying the conversion result by the kana-kanji conversion control circuit, and the conversion. In a kana-kanji conversion processing device having means for receiving finalized information among the results, part-of-speech information is stored in the word dictionary storage means.

Prefixes and suffixes are stored, and means is provided for storing the confirmed information in correspondence with input character strings, and in the output character column of the storage means, for a certain specified part of speech, this is stored as an output character. By recording part-of-speech codes instead of columns, it is possible to improve the automatic recognition rate of homophones and homologous words while reducing the burden on the operator, making it possible to create a practical kana-kanji conversion processing device. This has a remarkable effect.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing a part of an independent word dictionary, Fig. 2 is a block diagram of a kana-kanji conversion processing device which is an embodiment of the present invention, Figs. 15 is a diagram showing the character arrangement history of the keyboard device, FIG. 16 is a diagram showing the contents of the same sound phrase buffer, and FIG. 17 is a program flowchart. 11: Keyboard device, 12: Input character buffer register, 1
32 control circuit, 14: output editor, 15: display device, 1
6: Printer, 17: Connection matrix table, 18: Independent word dictionary,
19: Prefix table, 20: Suffix table, 21: Conjugation table,
22: Adjunct word table, 23: Hiragana katakana correspondence table, 24
: Output sentence register, 31: Homogeneous phrase buffer. Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Written amendment to the procedure (voluntary) February 28, 1980 Mr. Kazuo Wakasugi, Commissioner of the Patent Office, Case Description 1981 Special Fraud I No. 212478 No. 2 Name of the invention Kana-kanji conversion processing device 3 Relationship with the person making the amendment Patent applicant address Hide〇-E 1-3 Nakamagome, Ota-ku, Tokyo Representative: Takeshi Oue 4, Agent: 5 & Number of inventions to be increased by amendment
Shi0,? ? Target of amendment (1) The description in the "Detailed description of the invention" column is amended as follows. 1) tJ4D line 4D "Homophone buffer memory", r
liL syllable buffer memory”. 2) Correct everything from lines 17 to 20 as follows. ``Words consisting of a suffix, a numeral, a place name, etc. (these are called ``same words''), for example, Article 1, 2nd moxibustion,...; Yokohama city, Yozaki city, etc. are each individually Since the above-mentioned same-sound edicts are stored in the same-sound phrase buffer memory, there is a problem that the capacity of the same-sound phrase buffer memory becomes large even though they are used less frequently, and the negative phase of the operator required for selection also increases. . 3) Page 5, lines 6 to 6, ``while reducing the burden on the operator'' was changed to ``without increasing the capacity of the homophone phrase buffer memory.''
In order to reduce the burden on the operator as described above, the correction is made as d. 4) Change “(number word)” from page 14, line 20 to “(number word\first
(marked with ■ in Figure 6)". 5) On page M16, line 12, after "An example is shown in Fig. 16.", add "In Fig. 16, ■ indicates a place name hood." 6) On page 17, line δ to line 9, "while significantly reducing the burden on the operator" is corrected to "without increasing the capacity of the storage means and while reducing the burden on the operator." ■ Correct the O mark in Figure 16 of the drawing (○ indicated by the red line)
all additions). Figure 16

Claims (1)

[Claims] A means for inputting Japanese sentences as kana sentences, a means for temporarily storing character strings input by the input means, a word dictionary storage means including output ranking information of homophone and allograph words, and a search for the word dictionary storage means. a kana-kanji conversion control circuit for converting the contents of the temporary storage means into a sentence mixed with kanji; a means for displaying a conversion result by the kana-kanji conversion control circuit; and a means for receiving definitive information of the conversion result. In the kanji conversion processing device, part of speech information, prefixes, and suffixes are stored in the word dictionary storage means, and
Means for storing the confirmed information in correspondence with the input character string is provided, and an output character column of the storage means is configured to record a part-of-speech code for a certain specified part of speech instead of the output character string. This is a kana-kanji conversion processing device.