JPH08249331A - Document processing device - Google Patents

Abstract

(57) [Summary] [Purpose] Performs syntactic analysis, points out clauses whose affiliation is not uniquely determined, and moves clauses to appropriate positions or inserts commas. In addition, it is possible to point out a phrase that is not in the standard case element appearance order and move the phrase by using the case pattern dictionary. [Constitution] When the morpheme analysis means 6 performs morpheme analysis and the syntactic analysis means 7 performs syntactic analysis, and when the dependent check means 10 determines that the dependent is ambiguous, the phrase moving means 9 or the comma inserting means 11 makes it ambiguous. Reduce and eliminate sex.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a Japanese word processor, a machine translation device, etc. for proofreading / revising Japanese.

[0002]

2. Description of the Related Art Conventionally, there are few document processing devices that perform Japanese proofreading / reviewing that perform parsing processing that is always performed by Japanese-English machine translators, etc. Most of them are morphological analysis processing and point out errors. Most did.

For example, the invention disclosed in Japanese Patent Laid-Open No. 4-213155 describes that morphological analysis processing is performed and typographical error candidates are output only when a more probable word sequence exists. . If there is an unregistered word that has not been registered in the word dictionary and a homonym whose connection cost is lower than the threshold value, that is, grammatically, by using the cost table that defines the appearance probability of the surface layer from the word part of speech and other information. Therefore, the typographical error candidate is output only when a certain word sequence exists.

Further, JP-A-62-150461 and JP-A-62-150462
A pair of adverbs and auxiliary verbs having a sympathetic relationship is stored in, and detection / correction is performed when there is no corresponding combination expression. This makes it possible to detect / correct a portion that may have an error in the usage of words having a synergistic relationship.

In Japanese Patent Laid-Open No. 62-209659, morphological analysis and syntactic analysis are performed, and a grammar with a constraint condition is used.
By pointing out the part corresponding to the constraint condition,
Supports proofreading.

[0006]

However, when the calibration / correction is performed by performing only the morphological analysis, the dependency relation is ignored, so that the typographical error can be corrected, but the dependency relation is clear. It may not be possible to calibrate / correct.

For example, the following two sentences will be taken as an example.

(1) "He saw a rest." (2) "He saw a rest." In these two sentences, "he" and "rest"? "saw"
It can be ambiguous. Generally, in sentence (1), it is natural for "he" to look "far" in the distance, such as "he saw someone (resting) taking a break." ) Sentence, it is natural to understand that "(someone) saw him resting", such that "he" relates to a "resting" nearby.

Therefore, in the present invention, by performing a syntactic analysis, the dependency relation is analyzed, and when the dependency relation is ambiguous, the clause is moved to a position where the ambiguity is resolved,
By inserting a punctuation mark or a comma immediately after the dependency phrase of the dependency, the ambiguity of the dependency is resolved and the sentence is easy to read.

For example, in the resolution of ambiguity by moving clauses, (1a) "He saw a break.", And in resolution of ambiguity by reading points, (1b) "He took a break. I saw it doing. "

In this case, if there are more punctuation marks or commas than necessary, the sentence becomes rather difficult to read. Therefore, a syntactic analysis is performed and only ambiguity is inserted.

Also, in the case of parsing, instead of performing a special process of putting a constraint condition on the grammar as in the conventional technique, all the elements related to one idiom are recognized. In addition, case analysis is performed to compare with case patterns held as dictionary information, and word order is checked.

[0013]

[Means for Solving the Problems]

[Means of Claim 1] Editing means for inputting / editing Japanese, storage means for storing the Japanese sentence input by the editing means, display means for displaying the Japanese sentence, and dictionary and grammar. In a document processing device comprising a dictionary storage means, a morpheme analysis means for performing a morpheme analysis based on a dictionary or a grammar stored in the dictionary storage means, and a syntactic analysis means for syntactic analysis, Dependent destination checking means for determining the presence / absence of a clause whose destination is not uniquely determined, and when the dependent destination checking means has an ambiguous clause, moves the clause to a position where the ambiguity of the dependent party can be resolved. Means are provided.

[Means for Claim 2] Editing means for inputting / editing Japanese, storage means for storing the Japanese sentence input by the editing means, display means for displaying the Japanese sentence, dictionary and grammar In a document processing device comprising a dictionary storage means for storing, a morpheme analysis means for performing a morpheme analysis based on a dictionary or a grammar stored in the dictionary storage means, and a syntax analysis means for performing a syntactic analysis. If there is an ambiguous phrase in the dependency check means that determines whether there is a phrase whose dependency is not uniquely determined from the result, and if there is an ambiguous phrase in the dependency check, place a reading mark or comma at a position where the ambiguity of the dependency can be resolved. A comma insertion means for inserting is provided.

[Means for Claim 3] In the dictionary storage means,
The co-occurrence information of the bunsetsu and the case pattern of the noun are stored, and the affiliation destination checking unit refers to the co-occurrence information of the bunsetsu stored in the dictionary storage unit to check the affiliation destination.

[Means for Claim 4] In the dictionary storage means,
The case pattern of the idiom is stored, the result check unit compares the analysis result with the case pattern stored in the dictionary storage unit, and the input sentence is stored in the order of the case pattern stored in the dictionary storage unit. A phrase moving means for moving the noun phrase is provided.

[0017]

According to the first aspect of the present invention, the syntactic analysis unit is performed on the input sentence input by the editing unit,
The result is judged by the contact check means, and if there is an ambiguous phrase, that fact is displayed, and when the user writes the sentence in the meaning related to the phrase farther from the contact candidates, The bunsetsu moving means can efficiently move the bunsetsu to the position immediately before the received bunsetsu, without the user having to bother to specify the range or move the bunsetsu.

According to the second aspect of the present invention, the ambiguity is resolved by moving the clause in the above-mentioned claim 1, but the operation such as moving the cursor immediately after the related clause is performed by the comma inserting means. You can insert punctuation marks or commas without doing.

According to the third aspect of the invention, the co-occurrence information of the bunsetsu is used in the dependency check means of claim 1 and claim 2 by utilizing the co-occurrence information of the bunsetsu and the case pattern of the noun. By doing so, the dependency check is performed, the phrase is moved or the reading point / comma is inserted, and the ambiguity of the dependency is resolved.

According to the fourth aspect of the present invention, the dictionary storage means stores a case pattern of a noun, and the dependency check means compares the analysis result with the case pattern stored in the dictionary storage means. Then, the phrase movement means for moving the noun phrase clause of the input sentence is provided in the order of the case patterns stored in the dictionary storage means.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a document processing apparatus according to the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing an example of the configuration of a document processing apparatus according to the present invention. 1 is control means, 2 is input / editing means, 3 is display means, 4 is dictionary storage means, 5 is grammar storage means, 6 is morpheme analysis means, 7 is syntactic analysis means, 8 is input character storage means, and 9 is Phrase movement means, 10 is a check destination means, 11 is a comma insertion means, 12 is a phrase insertion means, 13 is a case pattern storage means,
Reference numeral 14 indicates a standard word order checking means.

Under the control of the control means 1, the input sentence input from the input means 2 is stored in the input character storage means 8. At the same time, this content is displayed on the display means. The morphological analysis unit 6 uses the dictionary of the dictionary storage unit 4 and the grammar of the grammar storage unit 5 to divide a sentence into phrases.

The syntactic analysis means 7 creates a syntactic analysis tree using the dictionary of the dictionary storage means 4, the case pattern of the case pattern storage means 13 and the grammar of the grammar storage means 5. From the syntactic analysis tree, the dependent check means 10 detects an ambiguous sentence related to the dependent, and the display means 3 displays the ambiguous sentence.
Also, the phrase moving unit 9 moves to a position where the ambiguity is resolved or reduced, and the comma insertion unit 11 inserts a reading point or a comma.

Further, the syntax analysis tree and case pattern storage unit 1
From the case pattern No. 3, the standard word order checking unit 14 checks whether or not they appear in the order of the case element standard, and the phrase moving unit 9 moves them.

FIG. 2 is a flow chart showing the processing of the present invention. FIG. 9 is a diagram showing an example of a screen display image. First, there is a main window that displays the text to be composed. If the full text cannot be displayed, it is possible to check the entire text by using a function such as scrolling.

In the present invention, since the processing is basically for each sentence, a window for displaying the sentence to be processed and the pointed out contents is prepared. In FIG. 9, it is shown as a “window showing the indication target sentence”. In FIG. 9A and FIG. 9B, the underlined clauses indicate the clauses that are better to be moved, and the blank frames indicate the positions to be moved.

A window in which the calibration operation is used as a menu is shown as a "calibration operation selection menu window".
"●" to select whether to follow the system's instructions
“O” indicates selection and non-selection, respectively, which can be changed / selected with a mouse or the like.

Although not shown in FIG. 9, a window for displaying the reason (content) of the problem may be displayed. For example, in the case of Fig. 9 (a), the display "There is an ambiguous phrase related to the person concerned. Are you sure you want to move the corresponding phrase to a position that can reduce the ambiguity?" The order of case particles for adjectives is different from the standard. Are you sure you want to change the standard word order? ”Is displayed.

FIG. 2 is a main flow. Step S1
Then, one input sentence is read into the input character storage means 8.
At the same time, a display is made on the display means 3. Step S
In 2, the morphological analysis unit 6 creates a clause using the dictionary of the dictionary storage unit 4 and the grammar of the grammar storage unit 5. In step S3, calibration support that can be performed at the morphological analysis level is provided. For example, it points out typographical errors and points out inconsistencies in word endings.

In step S4, it is judged whether the input sentence is changed or not based on the indication in the proofreading support performed in step S3, and if it is changed, the process proceeds to step S2 to perform the morphological analysis again. Return. If no change has been made, the process proceeds to step S6.

In step S6, a syntax analysis process is performed.
The syntax analysis algorithm is not special, but case analysis / semantic analysis using case patterns is performed.

In step S7, the calibration support is performed using the result of step S6. Details will be described with reference to the flows of FIGS. 3 and 4 and the example sentences of FIG. In addition, this FIG.
The order of the processes shown in FIG. 4 does not matter.

Next, with reference to FIG. 3, the processing when there is ambiguity in the dependency is shown. In step S11, as shown in FIG.
In the sentence like (1-3), the result of the syntactic analysis may have two types of dependency as shown in FIG. FIG. 7 shows a dependency structure, in which related words are arranged below the receiving word.
In Figure 7 (1-3a), "Book" and "Yesterday" are "Read",
Indicates that "buying" is related to "book", and (1-3b)
"Yesterday" is "buy", "buy" is "book",
It indicates that "reading a book" is related to "reading". FIG. 6 (1-3) shows this structure in the same manner. Using the contact check means 10, a phrase that may be at two places is searched from this structure, and "yesterday" is found. If there is a phrase that may be related to two or more places, the process returns to step S12, and if not, the process returns to the main flow.

In the case of the third aspect of the present invention, a case analysis and a semantic analysis are performed using the case pattern stored in the case pattern storage means 13 to determine which one of the two or more related parties is likely to be related. to decide.

In FIG. 5 (1-1a), as shown in FIG. 6 (1-1a), "playing the guitar" is changed to "playing" at a farther position.
-1b), as shown in FIG. 6 (1-1b), relates to a closer phrase “singing”. Similarly, in the examples of (1-2a) and (1-2b) of FIG. 5, the example of (1-2a) relates to a farther bunsetsu. According to claim 3, in this step, if the system determines that there is an ambiguous phrase related to the related item and that the related phrase is related to a farther one, the step S3
2. If not, return to the main flow.

That is, in the sentences (1-1a) and (1-2a) in FIG. 5, the process proceeds to step S12, and the sentences (1-1b) and (1-2b) in FIGS.
Statement returns to the main flow. In addition, in the example of (1-3), it cannot be determined whether "yesterday" relates to "buying" or "reading". In addition, "red" in Fig. 5 (1-4) and Fig. 6 (1-4).
It is not possible to judge whether is related to "company" or "car" (it can be judged if the knowledge that there is no "red company" is stored). In these cases as well, it is up to the user's judgment in step S12.

In step S12, the characteristic display (for example, color display, underline display, etc.) of the ambiguous phrase of the destination and the position of the destination are displayed. In the case of the sentence of (1-1a) in FIG. 5, "window showing the indicated sentence" in (1-1a) of FIG. 9
Is displayed. Also, selection items are displayed as in the “calibration operation menu window”. Here, the option of moving the phrase, inserting a reading point (or a comma), or leaving it as it is is displayed, and the process proceeds to step S13.

In step S13, the user selects from the display options in step S12, and the process proceeds to step S14.

In step S14, when the phrase movement is selected in step S13, the feature-displayed phrase is moved to the destination position. The destination is just before the clause of the destination. Also, when insertion of a reading point is selected, a reading point (or a comma) is inserted immediately after the phrase displayed as a feature. After each process is completed, the process returns to the main flow.

Next, the processing for checking the word order using case analysis and changing the word order will be described. In step S21, as a result of the syntactic analysis, it is analyzed what clauses are associated with which terms and in what order. By comparing this with the case pattern data, it is judged whether or not the word order is standard. For example, in FIGS. 5 (2-1) and (2-2),
The terms "written" include "book" and "me" in that order. In addition, in the verb "to regard", "correct answer" and "this" are related in this order. In this idiom,
Case pattern data is shown in FIG. If the order of each element of the case pattern is the standard word order, both are different. If it is different from the standard word order like this,
The process proceeds to step S22, and if they are not different, the process returns to the main flow.

In step S22, the phrase movement for displaying in the standard word order is displayed. That is, the phrase to be moved and the move destination are displayed as in the “window showing the indicated object sentence” in (2-1) of FIG. If there are three or more case elements, steps S21 to S24 are repeated.

Considering the change of word order as the movement of a clause that is a case element, the case element closest to the word in the standard word order is checked, and another case is inserted between the case element and the word in the input sentence. When the element is included, the destination is set just before the idiom. When it is repeated, each element next to the word is checked, and the destination is set immediately before the case element closest to the word that has finished moving first. After that, the process proceeds to step S23.

In step S23, the user selects whether to move. When moving, step S2
4. If not, return to the main flow.

In step S24, the feature-displayed clause is moved to the destination position. Then return to the main flow.

In step S7, the contents of the input character storage means 8 and the contents of the display means 3 are changed depending on the result of the calibration by the conventional calibration support function of step S3 and the result of the calibration by the calibration support function of the present invention of step S6. To be done.

If there is a next sentence in step S8, the process returns to step S1. If not, the process ends.

[0047]

【The invention's effect】

[Effect of Claim 1] According to the present invention, when a distant affiliation is related to an ambiguous bunsetsu, the bunsetsu is related to the bunsetsu which is not intended by the user. It is possible to move to the position immediately before and correct the disambiguation or the sentence to reduce the ambiguity, and it is possible to accurately convey the intention of the writer to the reader.

[Effect of claim 2] It is also implied that a punctuation mark or a comma is inserted immediately after the bunsetsu, instead of moving the bunsetsu, to indicate to the reader that the bunsetsu is distant. It is possible to convey and reduce ambiguity.

[Effect of claim 3] Further, by using the case analysis / co-occurrence information, it becomes possible to accurately determine the dependency.

[Effect of claim 4] In Japanese, the word order of the case elements of the idiom is relatively free, but the standard word order is determined to some extent except for the case of special effect.
Easy to reach the reader. Therefore, by pointing out a non-standard word order and moving to a standard position, it is easy for the reader to understand.

[Brief description of drawings]

FIG. 1 is a block diagram according to an embodiment of the present invention.

FIG. 2 is a main flow diagram according to the present invention.

FIG. 3 is a flowchart relating to the processing of claim 1 or 2 of the present invention.

FIG. 4 is a flow chart relating to the processing of claim 4 of the present invention.

FIG. 5 is a diagram showing an example of an input sentence.

FIG. 6 is a diagram showing a dependency structure of an example of an input sentence.

FIG. 7 is a diagram showing a result of syntax analysis of an example of an input sentence.

FIG. 8 is a diagram showing an example of a case pattern of a vocabulary dictionary.

FIG. 9 is a diagram showing a screen display example according to an embodiment of the present invention.

[Explanation of symbols]

 1 control means 2 input / insertion means 3 display means 4 dictionary storage means 5 grammar storage means 6 morpheme analysis means 7 syntactic analysis means 8 input character storage means 9 phrase movement means 10 reference destination check means 11 comma insertion means 12 phrase insertion means 13 Case pattern storage means 14 Standard word order check means

Claims (4)

[Claims] 1. An editing unit for inputting / editing Japanese, a storage unit for storing the Japanese sentence input by the editing unit, a display unit for displaying the Japanese sentence, and a dictionary storage for storing a dictionary and grammar. Means, a morpheme analysis means for performing morpheme analysis based on the dictionary or grammar stored in the dictionary storage means, and a document processing device comprising syntactic analysis means for syntactic analysis. Dependent check means for judging the presence or absence of a clause that cannot be uniquely determined, and clause moving means for moving the clause to a position where the ambiguity of the dependent party can be resolved when there is an ambiguous clause in the dependent check means. A document processing device comprising: 2. An editing unit for inputting / editing Japanese, a storage unit for storing the Japanese sentence input by the editing unit, a display unit for displaying the Japanese sentence, and a dictionary storage for storing a dictionary and grammar. Means, a morpheme analysis means for performing morpheme analysis based on the dictionary or grammar stored in the dictionary storage means, and a document processing device comprising syntactic analysis means for syntactic analysis. Dependent check means for determining the presence or absence of a clause that cannot be uniquely determined, and, if there is an ambiguous clause in the dependent check means, a comma insertion means for inserting a reading point or a comma at a position where the ambiguity of the dependee can be resolved. A document processing apparatus comprising: 3. The lexical co-occurrence information and the noun case pattern are stored in the dictionary storage means, and the affiliation destination checking means refers to the bunsetsu co-occurrence information stored in the dictionary storage means. The document processing apparatus according to claim 1 or 2, wherein a check is made on the person in charge. 4. A case pattern of a noun is stored in the dictionary storage means, and the relationship check means compares the analysis result with the case pattern stored in the dictionary storage means and stores it in the dictionary storage means. A document processing apparatus comprising: a phrase moving unit that moves a noun phrase phrase of an input sentence in the order of the generated case patterns.