JPH061468B2 - Japanese sentence proofreading device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial field of use> The present invention relates to a Japanese sentence proofreading device for extracting and proofreading notations and expressions that do not meet specific criteria from Japanese sentences.

<Conventional Technology> In Europe and the United States, word processors have developed from an early stage, and related technologies have advanced, and devices with spell check and collect functions have been put into practical use.

Compared to Western words that are written in units of words, Japanese is usually written in Kanji and Kana interchange without punctuation, and the orthography is not thorough. However, a device for automating calibration has not been put into practical use.

Conventionally used as a proofreading method for Japanese, multiple people make a pair and read together to extract problematic points, or a person with specialized knowledge of proofreading sequentially adds and verifies the original sentence. Is the way.

For the former, there has been proposed a proofreading device capable of analyzing a sentence in which kanji and kana are mixed, converting it into a voice, and performing reading together with a machine.

However, the latter requires the function of supporting the automation of calibration or the specialization of calibration, and requires a higher level of technology. Therefore, a means for configuring the device by only proposing a concept,
No specific means of calibration has been reported.

<Problems to be solved by the invention> To carry out calibration, specialized knowledge of calibration is required.
It took time to learn it. In addition, the number of people who have specialized knowledge of proofreading is small, and there are problems when proofreading a large amount of sentences or when urgent proofreading is required.

Also, from the standpoint of an expert who has the knowledge of calibration, it is necessary to pay attention to the fine details in order to perform calibration correctly.
I had a mental burden.

Nevertheless, there were some errors that were difficult for humans to find and errors that were easily overlooked, and there was a problem in terms of calibration accuracy.

In addition, when an ordinary person who does not have specialized knowledge of proofreading creates a document using a word processor, etc., reprocessing such as reading, printing, and copying of the document file is necessary for proofreading small mistakes. There was a problem in the efficiency of document creation work.

The Western and American spell check and collect functions are aimed at proofreading notations and terms, and there was no proofreading device to proofread syntactic and homophone errors.

The present invention performs morphological analysis and syntactic analysis to identify a character / symbol string to be proofread from a Japanese sentence (hereinafter, the character / symbol string may be referred to as a character string without special notice). By extracting and presenting the information, the user of the apparatus is assisted in the calibration work, and the problem is solved.

<Means for Solving Problems> According to the present invention, a dictionary means that stores a headline and morpheme information of the headline, a grammar means that stores Japanese grammatical information, a syntax generation rule, and a constraint of the generation rule. For each of the temporary character strings divided by the dividing means, a syntax rule means for storing the condition, a storage means for storing the input sentence, a dividing means for temporarily dividing the sentence stored in the storage means, A morpheme analysis means for performing a morpheme analysis using the dictionary means and the grammar means, a storage means for storing morpheme information when the morpheme analysis by the morpheme analysis means is successful, and a morpheme information stored in the storage means The parsing means for applying the generation rule of the syntactic rule means to generate a syntactic parse tree and the constraint condition of the syntactic rule means, and the syntactic solution generated by the syntactic analyzing means. Display means for judging the syntactic validity of the parsing tree and displaying a syntactic component having a low syntactic validity as a character string to be proofread separately from the other, and a character displayed by the display means to distinguish from the others And a key input changing means for changing the column by key input.

<Operation> The present invention temporarily divides an input sentence, morphologically analyzes each divided temporary character string using a dictionary means and a grammar means, and accumulates morpheme information when the morpheme analysis is successful. Then, the generation rule of the syntactic rule means is applied to the accumulated morpheme information to generate a syntactic parse tree, the syntactic validity of the generated syntactic parse tree is referred to by referring to the constraint condition of the syntactic rule means. The sentence is proofread by determining the above, displaying a syntactic component having a low syntactical validity as a character string to be proofread separately from the others, and changing the character string by key input.

<Example> The present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram of a Japanese sentence proofreading device according to the present invention.

In the figure, 1 is a means for inputting / editing a Japanese character string. Reference numeral 2 is a means for storing the Japanese character string input by the input means. Although a keyboard is usually used as the input means, an external storage means for storing the input Japanese character string, such as a floppy disk or a magnetic tape, can be used instead of the sequential input. That is, there may be a configuration in which 1 is omitted.

Reference numeral 3 is a means for storing a dictionary consisting of Japanese character strings that meet the criteria. 4 is a means for storing Japanese grammar. Since the Japanese orthography has not been established, it is said that the notation, grammar and expressions that meet certain standards are correct. Hereinafter, when the correct notation, grammar, and expression are described, it means that they meet a certain standard.

Reference numeral 5 is a control means for controlling input / output, collation, information storage, display and the like. The control means includes a means for storing the result obtained by the control, and 6 is a means for parsing Japanese. 7 is the input Japanese, the result of the verification process,
It is a display means such as a CRT that displays a character string to be calibrated. Reference numeral 8 is a means for performing morphological analysis using a dictionary, grammar and the like.

FIG. 2 shows a string of character strings input by the input means.
In this case, it indicates that "ha" has been input extra. "Ha" has both the word "ha" and the particle "ha" that expresses victory within the limits of common kanji, such as competing for victory.

FIG. 3 is an example of the dictionary described in 3. Reference numeral 9 is a headline to be compared with the input character string. Reference numeral 10 is one of the morpheme information of the headline.

FIG. 4 is an example of the grammatical information stored in 4. The heading shown in FIG. 3 is also shown for the sake of clarity. 1
Reference numeral 1 is information that defines the connection relation of each morpheme information. For example, a pronoun indicates that a case particle can be connected to a particle. Reference numeral 12 is information that defines whether each morpheme information can satisfy the phrase ending condition, and ○ indicates that the phrase ending condition is satisfied, and × indicates that it is not satisfied.

FIG. 5 is a correspondence table of morpheme information and grammar category names.
13 is an example of a grammar category name. Reference numeral 14 is morpheme information corresponding to 13. A grammar category name in which 14 is blank indicates that it is not a terminator.

FIG. 6 is an example of syntax rules. Reference numeral 15 is a syntax generation rule. 16 represents a constraint condition when 15 is applied.

FIG. 7 is a schematic flow chart of the present invention. 17 stores the input character string in the stack in 5. 18 tentatively divides the accumulated character / symbol string by a kanji code or the like. Reference numeral 19 reads each provisionally divided input character string. 20 performs morphological analysis of the input character string read in 19 using a dictionary and grammar. Depending on the result, the process branches if the morphological analysis succeeds or fails. Reference numeral 21 stores morpheme information that has been successfully analyzed.

22 receives the morpheme information obtained in 21 as an input and applies a syntax rule to generate a syntax analysis tree. 23 collates 16 against a set of syntactic rules having a parse tree grown to SENT to determine syntactic validity. 24 extracts syntactic elements having low syntactical validity as a character string to be proofread,
It is displayed separately from the others. 25 is proofreading based on the information of 24 by using a selecting means or a Japanese input / editing means. 26 examines the next input string to be matched.

Reference numeral 27 carries out a process when collation of either the dictionary or the grammar fails. The processing of 27 is outside the scope of this patent and will not be described in detail.

The actual processing flow will be described below using the example sentences shown in FIG. 2 and the dictionaries, grammars, and syntax rules shown in FIGS. 3, 4, 5, and 6.

First, the input string "He goes to school" is stacked at 5. Next, the input character string is provisionally divided using the code of the stacked character strings. For example, split before a character string that changes from hiragana to a code of another character type.

As a result, "he goes to school" means "he is" and "go to school."
And "Go." First, the processing of the morpheme of "he is ga" will be described. "Hi ga ga" is read by the control device 5 and collated with the dictionary headings in FIGS. In the present case, the part of "he" in the dictionary matches. This information is stacked in 5 and the connection condition of the corresponding heading is compared with the next input character string. In the present case, both are "ga" and they match. Next, the clause end condition of the item headed by "ga" in FIG. 3 is checked in 12 in FIG. From 12, it is found that the phrase ending condition is satisfied in the present case, and "he is" is stacked as a normal phrase at 5.

Next, the morpheme processing of “ha” is performed in the same procedure as “he is”.
As a result, "he is" and "ha" are put together as a sentence and "he is"
However, if he has two clauses, “he is” and “ha,” it is possible to satisfy the morphological condition.

Similarly, the morpheme processing of "go to school" and "go." Is performed.

Based on the above, the morpheme information "he ・ ・ pronoun", "ga ・ ・ case particle", "ha ・ ・ noun", "school ・ noun", "he ・ ・ case particle", "go. Form "is obtained.

Next, by referring to the correspondence table (FIG. 5) between morphemes and grammatical category names, "pronoun ... N", "case particle ... CP",
Get "Noun ... N" etc.

Next, by applying the syntax rules “NP → N” and “NPC → NP, CP” in the order of the sentence sequence according to FIG.
Summarize with NPC. By referring to the constraint conditions at the time of syntax generation,
In this case, NPCs are unconditionally put together.

Next, proceed to the "ha" and "school" parts that contain the problem. In the same way as above, "ha school" can be summarized as NP by "NP → N, N". In this case, "ha" corresponds to the constraint condition when referring to the constraint condition on the syntax generation.

By repeating the above procedure, the syntax analysis tree of FIG. 8 is obtained.

For the growth of the syntax tree, the part that contains the constraint condition is likely to contain the syntactic problem for the part that does not.
From the result of the syntactic analysis accumulated in 5, it is possible to extract a character string having a constraint on the syntactic growth. The text can be proofread by displaying it separately from the others. The constraint expression can be expressed by the morpheme itself, or can be specified by the category name of the morpheme.
Further, the criterion for judging the calibration character string from the constraint condition may be set by setting the constraint condition by a numerical value or a level value and comparing it with the level value.

The processing of the branch that does not appear in the above description will be briefly described. Two
6 is a module for selecting a morpheme or a character string to be syntactically analyzed. If there is no input character string, the series of processing ends. Reference numeral 27 is a process including a character string having a problem in morphological analysis, and the proofreading can be performed by a method different from that of the present patent.

<Effect of the Invention> The effect of the present invention is that since the expert knowledge of the calibration is built in, it is not necessary to acquire the expert knowledge, and even an amateur can perform the calibration with high accuracy. Since the main work of the calibration can be performed by this device, the user can reduce the working time of the calibration work.

In addition, it is possible to automatically extract a character string that needs proofreading from a Japanese sentence and present the character string to be replaced using a proofreading dictionary or display it separately so that the proofreader's Effective in reducing fatigue and mental burden.

In addition, it is possible to extract mistakes that are often overlooked by humans, such as the mistakes of katakana “f” and hiragana “he”, which is effective in improving the accuracy of calibration. Also,
By combining with a device such as a word processor,
A series of re-work such as printing and copying caused by proofreading can be omitted, and the efficiency of document creation can be expected to be improved.

In addition, it is possible to take out the part of the syntax and semantic errors, and it can be expected to improve the efficiency of the proofreading work.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of the device of the present invention, FIG. 2 is a diagram showing an example of a character string input by input means, FIG. 3 is a diagram showing an example of a dictionary, and FIG. 4 is an example of grammatical information. FIG. 5, FIG. 5 is a diagram showing an example of a proofreading dictionary, FIG. 6 is a schematic flow chart of processing, FIG. 7 is a diagram showing a display example of proofreading candidates, and FIG. It is a figure which shows the example applied in order. 1 is input means, 2 is input character storage means, 3 is dictionary storage means, 4 is grammar storage means, 5 is control means, 6 is syntactic analysis means, 7 is display means, and 8 is morpheme analysis means.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Mikio Osaki 22-22 Nagaike-cho, Abeno-ku, Osaka-shi, Osaka Inside Sharp Corporation (72) Inventor Taro Morishita 22-22 Nagaike-cho, Abeno-ku, Osaka Within the corporation

Claims (1)

[Claims] 1. A dictionary means storing a headline and morpheme information of the headline, a grammar means storing Japanese grammatical information, a syntax rule means storing a syntax generation rule and constraint conditions of the generation rule, A storage unit that stores the input sentence, a dividing unit that temporarily divides the sentence stored in the storage unit, and the dictionary unit and the grammar unit for each temporary character string divided by the dividing unit. Morphological analysis means for using morpheme analysis, accumulation means for accumulating morpheme information when the morpheme analysis by the morpheme analysis means is successful, generation rules of the syntax rule means for the morpheme information accumulated in the accumulation means By applying it, the syntactic analysis means for generating the syntactic analysis tree and the syntactical validity of the syntactic analysis tree generated by the syntactic analysis means are determined by referring to the constraint conditions of the syntactic rule means. Display means for distinguishing and displaying a syntax component having a low syntactical validity as a character string to be proofread, and key input for changing the character string displayed separately from the other by the display means by key input. A Japanese grammar proofreading device comprising a changing means.