JPH06243163A - Document analyzing device - Google Patents

Abstract

PURPOSE:To lower possibility that an error is generated in recognizing an auxiliary verb scope, and to efficiently execute a processing in a document processor (machine translation device, etc.). CONSTITUTION:An auxiliary verb analyzing part 52 is called out at the time when an analysis related to a word being equivalent to an auxiliary verb is executed in an analytic process by an analyzing part 51, detects a scope level of a YOGEN (inflective words such as a verb, an adjective, etc.) candidate in an auxiliary verb scope and a declinable word candidate in the auxiliary verb scope concerned in accordance with a processing rule based on an appearance pattern, etc., of a word being equivalent to an auxiliary verb and a word being equivalent to a declinable word, and executes recognition of the auxiliary verb scope, based on the detection concerned. At the time of correcting an output sentence (translated sentence) of a machine translation device, an analysis result display means 6 displays the declinable word candidate in the auxiliary verb scope detected by the auxiliary verb analyzing part 52 and information related to the auxiliary verb scope for showing the scope level, etc. An auxiliary verb shared designating input means 7 changes the contents of the auxiliary verb scope recognized by the auxiliary verb analyzing part 52, based on designation from a user to the information related to the auxiliary verb scope.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document analysis device applied to a document processing device such as a machine translation device and a document proofreading device, and particularly to a sentence description in a language including an auxiliary verb equivalent word and a noun equivalent word. The present invention relates to a document analysis device that analyzes.

[0002]

2. Description of the Related Art Many languages (natural languages) have auxiliary verb equivalents and verb equivalents. Here, the verb equivalent word is a word or word group that can be used alone as a predicate. Further, the auxiliary verb equivalent word is a word or a group of words to which an additional meaning such as law, tense, and statement is added to a noun equivalent word that has a central meaning of a predicate.
The auxiliary verb equivalent word and the synonym equivalent word are not necessarily limited to the words indicating the auxiliary verb and the synonym in the grammar of each language, and the information processing system (document analysis device as a word or a word group having the above functions The information processing system to which this belongs) is also included.

In a sentence described in a language in which the auxiliary verb equivalent word and the adjective equivalent word exist, one auxiliary verb equivalent word is applied to a plurality of the adjective equivalent words (a plurality of the equivalent adjective words is 1). There are cases in which one auxiliary verb equivalent word is shared), and the auxiliary verb equivalent word gives an additional meaning in common to those verb equivalent words. Here, the range of the adjective equivalent words commonly applied to a certain auxiliary verb equivalent word (information indicating the group of adjective equivalent words sharing the auxiliary verb equivalent word) is called the "auxiliary verb scope" of the auxiliary verb equivalent word. , Each synonym word belonging to the auxiliary verb scope is referred to as "individual verb within the auxiliary verb scope".

Conventionally, in a document analysis device, when an auxiliary verb equivalent word is detected in the process of sentence analysis (morphological analysis, syntactic analysis, semantic analysis, etc.), the auxiliary verb scope of the auxiliary verb equivalent word is certified. There is.

In the conventional document analysis device, the qualification of the auxiliary verb scope depends only on the processing rule (the processing rule based on the appearance pattern of the auxiliary verb equivalent word and the noun equivalent word) preset in the document analysis device. Was being done. However, in sentences written in natural language, it is difficult to clearly understand the regularity of the auxiliary verb scope. Therefore, proper processing rules regarding the recognition of the auxiliary verb scope have not been completely established. Therefore, if an attempt is made to certify an auxiliary verb scope that depends only on processing rules, there is a high possibility that the erroneous certification will occur. That is, the auxiliary verb equivalent word is certified so that the auxiliary verb equivalent word is not applied to any of the plurality of adjective equivalent words that should be shared, or the auxiliary verb equivalent word is shared by an excessive number of equivalent word equivalents. Certification was sometimes done.

Further, in the conventional document analysis device, such an error in recognizing the auxiliary verb scope cannot be directly corrected by the user of the document processing device to which the document analysis device is applied. That is, the conventional document analysis device
Neither did they display the analysis result having the contents of the auxiliary verb scope certification, nor did they accept the specification of the correction from the user.

[0007]

In the conventional document analysis device, the qualification of the auxiliary verb scope depends only on the processing rule set in advance in the document analysis device, and the user cannot recognize the erroneous recognition of the auxiliary verb scope. Since it cannot be corrected directly, there is a high possibility that an error in the recognition of the auxiliary verb scope will occur, and there will be time and effort to correct the error after the processing of the document processing device to which the document analysis device is applied. There is a drawback in that the processing efficiency of the device becomes poor.

[0008] For example, when an error of this kind occurs in a conventional document analysis device applied to a machine translation device, a translation or an extra phrase in which the auxiliary verb equivalent word is not applied to the auxiliary verb equivalent word should be applied. Inappropriate translations, such as translations in which the equivalent verb is applied to the equivalent verb, appear, resulting in syntactic destruction or semantic destruction. As a result, the user needs to recorrect the output sentence (translated sentence) of the machine translation device, and a lot of trouble is required for the correction.

As a conventional technique for an auxiliary verb (auxiliary verb equivalent) in a document analysis device, for example, it is determined whether the Japanese auxiliary verb "ru" or "ru" has a function of passive or respect. "Japanese sentence processing method (Japanese Patent Laid-Open No. 1-233661)" has been announced. However, there is no direct relationship between this conventional technology and the technology related to “accrediting auxiliary verb scope”,
The above-mentioned drawbacks could not be solved even by the analysis by this conventional technique.

In view of the above points, an object of the present invention is to carry out certification with a certain range in certification of the auxiliary verb scope (certification in consideration of some possibilities), and also reflect designation by the user. By performing the authorization, the possibility that an error occurs in the authorization of the auxiliary verb scope is reduced, and the document processing apparatus (the document processing apparatus to which the document analysis apparatus of the present invention is applied) is provided.
It is to provide a document analysis device capable of improving the efficiency of the processing in.

[0011]

A document analysis apparatus according to the present invention includes an analysis unit in an analysis unit for performing analysis such as morphological analysis, syntactic analysis and semantic analysis on an input sentence of a document processing apparatus, and an analysis by this analysis unit. Called when parsing the auxiliary verb equivalent word in the process of, the verb level in the auxiliary verb scope and the candidate verb in the auxiliary verb scope are determined according to the processing rule based on the appearance pattern of the auxiliary verb equivalent word and the phrase equivalent word. Auxiliary verb analysis unit in the analysis means for detecting and recognizing the auxiliary verb scope based on the detection, and the auxiliary verb scope noun candidate and scope level detected by the auxiliary verb analysis unit when correcting the output sentence of the document processing device. The analysis result display means for displaying the auxiliary verb scope-related information to be displayed, and the auxiliary verb score displayed by the analysis result display means. And a auxiliary verb shared specification input means for changing the contents of the certified auxiliary verb scope by the auxiliary verb analyzer based on the designation from the user for the relevant information.

[0012]

The present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a machine translation device including a document analysis device according to an embodiment of the present invention. The document analysis device of this embodiment is applied to a machine translation device that performs Japanese-English machine translation. The document analysis device of the present invention is a document processing device (machine translation device and document proofreading device) in which a sentence described by a language having an auxiliary verb equivalent word that may be applied to a plurality of verb equivalent words is a processing target sentence. Device, etc.)
The application is not limited to the machine translation device that performs Japanese-English machine translation.

The machine translation apparatus to which the document analysis apparatus of this embodiment is applied is an input unit 1, a dictionary 2, a generation unit 3, and
The output unit 4 and the document analysis device of this embodiment are included.

The document analysis apparatus of this embodiment comprises an analysis unit 5 having an analysis unit 51 and an auxiliary verb analysis unit 52, an analysis result display unit 6, and an auxiliary verb sharing designation input unit 7.

2 to 7 are diagrams for explaining the specific operation of the document analysis apparatus of this embodiment.

Next, the operation of the machine translation apparatus including the document analysis apparatus of this embodiment will be described. Here, a specific operation will be described with reference to FIGS.

First, the operation at the time of generation of a translated sentence (output sentence of the machine translation device) by the machine translation device (at the time of generation of the translated sentence at the stage where the designation from the user is not input) will be described.

The input means 1 inputs an original sentence (sentence to be machine translated) which is an input sentence of the machine translation device.

The analysis unit 51 in the analysis means 5 refers to the information described in the dictionary 2 and analyzes the original sentence (morphological analysis, syntactic analysis, semantic analysis, etc.) on the basis of preset analysis rules. I do. That is, the analysis unit 5
The reference numeral 1 analyzes the original sentence received from the input means 1 while referring to the dictionary 2 (while performing a search on the dictionary 2) and performs analysis based on the analysis rule to divide words in the original sentence, parts of speech and meaning of each word, and An analysis result indicating the relationship between words is generated.

The dictionary 2 includes information necessary for machine translation (grammatical attributes such as translated words and part-of-speech, and semantic classification such as semantic classification) for each headword in the original language (language describing the original sentence). Information such as attributes) is stored. By the way, even in a document processing apparatus other than the machine translation apparatus, there is a dictionary in which information necessary for each document processing is stored. For example, in the dictionary in the document proofreading apparatus, there is information other than the translated word in the information in the above-mentioned dictionary 2.

The auxiliary verb analysis unit 52 is called by the analysis unit 51 when the analysis of the auxiliary verb equivalent word is performed in the analysis process of the analysis unit 51.

The auxiliary verb analysis unit 52 follows a processing rule based on the appearance pattern of the auxiliary verb equivalent word and the adjective equivalent word, and the like. ) And information about the degree to which the candidate in the auxiliary verb scope can be a candidate in the auxiliary verb scope (hereinafter, “scope level”).
And), and the auxiliary verb scope is certified based on the detection. Here, the auxiliary verb analysis unit 52 in the present embodiment handles the scope level indicating the degree of possibility of being an auxiliary verb in-scope phrase in two stages (“high” and “low”) (scope level is 2). Needless to say, it is not limited to stage information). In addition, the auxiliary verb analysis unit 52 in the present embodiment certifies the auxiliary verb scope so that only the auxiliary verb scope candidate words having the scope level “high” are included in the auxiliary verb scope (certification of auxiliary verb scope). Is not limited to such a form of recognition).

The generating means 3 includes the analysis result generated by the analysis unit 51 (the auxiliary verb scope noun candidates and scope levels detected by the auxiliary verb analysis unit 52 and the auxiliary verb scope certified by the auxiliary verb analysis unit 52). Existing), a preset generation rule (rule for generating a translated sentence), and information described in the dictionary 2 to generate an appropriate translated sentence with respect to grammar and the like. Note that the generation means in the document processing device other than the machine translation device also generates an output sentence based on the analysis result by the analysis means, the generation rule (rule for generating the output sentence), and the information described in the dictionary. . For example, the generation means in the document proofreading device generates a proofread sentence based on a generation rule or the like.

The output means 4 receives the translated sentence generated by the generating means 3 and outputs the translated sentence (display on a display device or the like).

In the course of the above-described series of processing, the original sentence which is the analysis target of the analyzing unit 5, the analysis result generated by the analyzing unit 5, and the translated sentence generated by the generating unit 3 are stored in the machine translation device. Is stored in the storage area of and is referred to at the time of "correction of translated text" described later.

Secondly, the operation at the time of correcting the translated text by the machine translation device (when the translated text is generated at the stage of inputting the designation from the user and reflecting the designation) will be described.

When the user refers to the translated sentence output by the output means 4 and recognizes a sentence in which the auxiliary verb scope is erroneously recognized (hereinafter referred to as "auxiliary verb scope misidentified sentence"), the analysis result display means is displayed. 6 is requested to display the auxiliary verb scope misidentified sentence (both the original sentence and the translated sentence).

The analysis result display means 6 displays the auxiliary verb scope erroneous sentence on the display device.

A first example of this display is shown in FIG. 2 (the display of FIG. 2 also includes the display of auxiliary verb scope related information which will be described later). Note that this example is an example in which there is an error in the recognition of the user and there is actually no error in the recognition of the auxiliary verb scope.

In the example shown in FIG. 2, as the auxiliary verb scope erroneous sentence, the original sentence "This file can be modified and deleted." And the corresponding "This file"
can be modified and dele
ted. The translated text "" is displayed.

The auxiliary verb analysis unit 52, when analyzing the original sentence in FIG. 2, produces the appearance patterns of the auxiliary verb equivalent words and the synonym equivalent words (for example, "two synonym equivalent words connected by a reading point"). Pattern is shared with the auxiliary verbs that appear immediately after them, etc.), etc., and the auxiliary verb equivalents in the original sentence are “possible” (“possible” in Japanese grammar is independent. , Which is defined as an auxiliary verb equivalent in this machine translation device), is a verb (one of the equivalents) of the verb (one of the equivalents) and is deleted as an auxiliary verb scope candidate. "Delete"), determine (detect) their scope level as "high", and based on their detection, the "possible" auxiliary verb scope includes "correct" and "delete" I had done the certification. Such accreditation is based on the analysis results corresponding to this original text (see Fig. 3).
FIG. 3 is a diagram showing an analysis result of the original sentence in FIG. 2 generated by the analysis unit 51). The verbs “correct” and “delete” are called “{auxiliary verb. {Possible}}”. This is indicated by the addition of a marker (in the marker {A. {B}} in FIGS. 3 and 6, A indicates an attribute name and B indicates an attribute value).

Further, when the translation means in FIG. 2 is produced, the producing means 4 produces the translated word "be modif" for "correct" based on a predetermined production rule or the like.
The translations for “ied” and “delete” are “delete”
The translated word "can" for "possible" was translated to a position that is both "d (be deleted)" (the position immediately before both translated words).

The display shown in FIG. 2 reflects the above analysis and generation.

A second example of the above-mentioned display (display of auxiliary verb scope misidentified sentence by analysis result display means 6) is shown in FIG. 4 (the display of FIG. 4 also includes display of auxiliary verb scope-related information which will be described later). ). Note that this example is an example in which the auxiliary verb scope of the auxiliary verb equivalent word “nothing (none)” is actually incorrect.

In the example shown in FIG. 4, the auxiliary verb scope erroneous sentence is "remove the cover, adjust screws, clean,
Do not refuel. And the corresponding "Adjust the screen and do"
n't clean up or reference it
with removing the cover
r. The translated text "" is displayed.

Note that the auxiliary verb analysis unit 52, when analyzing the original sentence in FIG. 4, removes (removes (removes) the verb as a candidate for the auxiliary verb scope inflection for the auxiliary verb equivalent word “nothing”. ) ”,“ Adjust (adjust) ”,“ Clean (clean) ”and“ Refuel (refuel) ”are detected, and the scope level of“ Remove ”and“ Adjust ”is determined to be“ Low ” (Detect), determine (detect) the scope level of “clean” and “refuel” as “high”, and based on those detections, “clean” and “refuel” the auxiliary verb scope of “not” Was certified to include.

Further, when the translated text in FIG. 4 is created, the creating means 4 applies to the translated words "clean up" and "refuel" of "clean" and "refuel", but "removes". And the word "remo" that translates to
"do", which is a translation of "no", is placed at a position that does not overlap with "ving (remove)" and "adjust".
He was translating "n't (not)".

The display shown in FIG. 4 reflects the above analysis and generation.

As can be seen from the example of FIG. 4, it is difficult to properly recognize all of the auxiliary verb scopes only by the analysis by the auxiliary verb analysis unit 52 that depends only on the processing rules, and the auxiliary verb equivalent words should be applied to the verb equivalent words to be applied. There is a case in which an authorization that does not apply (for example, an authorization that "does not apply" to "adjust" in FIG. 4) or an auxiliary verb equivalent is applied to a noun equivalent that should not be applied. That is, there may be an excess or deficiency in the recognition of the auxiliary verb scope (this is a problem in the related art).

The user indicates to the analysis result display means 6 that
Specify the auxiliary verb equivalent word to be processed in the display shown in FIG. 2 or FIG. 4 (the display shown in FIG. 2 does not actually have an error, so the following explanation will be given for the display shown in FIG. 4) To do. This designation is a designation that prompts the display of information indicating the auxiliary verb scope noun candidate for the auxiliary verb equivalent word and its scope level (hereinafter referred to as “auxiliary verb scope related information”).

The analysis result display means 6 displays the auxiliary verb scope related information of the auxiliary verb equivalent word designated by the user. The auxiliary verb scope-related information in this embodiment is displayed with underlines in auxiliary verb scope misidentified sentences as shown in FIGS. 2 and 4 and the candidate list shown in FIG. 5 (in the auxiliary verb scope phrase for a specified auxiliary verb equivalent word). A list display having a list of candidates and an area (a square area in FIG. 5) for designating whether or not each auxiliary verb scope inflection candidate is an auxiliary verb scope inflection. To do. 2 and 4, the double solid line is attached to the designated auxiliary verb equivalent word, and the solid line is the auxiliary verb scope noun candidate corresponding to the auxiliary verb equivalent word, and the scope level is determined to be "high". The broken line is attached to the auxiliary verb candidate in the scope corresponding to the auxiliary verb equivalent word, and the scope level of which is determined to be "low". It goes without saying that the display mode of the auxiliary verb scope-related information is not limited to that shown in FIGS. 4 and 5.

For example, when the user specifies the auxiliary verb equivalent word "do not exist" for the auxiliary verb scope misidentification sentence shown in FIG. 4, the analysis result display means 6 displays the following.

For the auxiliary verb scope erroneous sentence, a double solid line is added to the underline of “nothing (none)”, and a solid line is added to the underline of “cleaning (cleaning)” and “refueling”. Then, a broken line is added to the underlines of “remove (remove)” and “adjust (adjust)” (see FIG. 4).

A candidate list as shown in FIG. 5 is displayed. In other words, each candidate in the auxiliary verb scope is displayed,
A square area, which is an area for indicating "whether or not each auxiliary verb scope candidate is an auxiliary verb scope word", is displayed corresponding to each auxiliary verb scope word candidate. The area corresponding to “cleaning” and “refueling” that are determined to be “in-scope verbs” is displayed in black.

The user refers to the above-mentioned auxiliary verb scope-related information and makes a designation for changing the auxiliary verb scope for a square area in the candidate list.

When the user makes an instruction to change the auxiliary verb scope, the auxiliary verb sharing designation input means 7 first receives the auxiliary verb analysis unit 52 so as to comply with the designation.
Change the contents of the auxiliary verb scope certified by.

For example, in the case where the user specifies that “adjust” or “becomes a verb within the auxiliary verb scope” in the candidate list shown in FIG. 5, the auxiliary verb sharing designation input means 7 is provided. Changes the contents of the auxiliary verb scope so that it conforms to the designation ("Adjust (adjust)", "Clean (clean)" and " Refuel (refuel) "is included).

When re-translation is instructed to the input means 1 after the change of the auxiliary verb scope as described above, the analysis means 5 causes the analysis result as shown in FIG. 6 based on the changed auxiliary verb scope. (The analysis result in which the mark “{auxiliary verb. {Negative}}” corresponding to the auxiliary verb equivalent “not” is also added to the verb “adjust”) is generated.

The generating means 3 generates a corrected translated sentence by using such an analysis result. The translated text after modification of the translated text in FIG. 4 is the translated text shown in FIG. 7.

The output unit 4 outputs (displays on the display device, etc.) the translated text after such correction.

[0052]

As described above, according to the present invention, the accreditation of the auxiliary verb scope is accredited with a certain range (accreditation considering some possibilities), and the designation from the user is also reflected. By performing such certification (correctly modifying (correcting) the excess and deficiency of the auxiliary verb scope based on the specification from the user), it is possible to reduce the possibility of error in the authorization of the auxiliary verb scope, and to reduce the error. (Document processing device to which the document analysis device of the present invention is applied) has the effect of improving the efficiency of processing.

For example, in the machine translation device to which the document analysis device of the present invention is applied, it is possible to reduce the correction man-hours of the translated sentence having the translated word corresponding to the auxiliary verb equivalent word.

Further, for example, in the document proofreading apparatus to which the document analyzing apparatus of the present invention is applied, a proofreading operation for automatically supplementing the auxiliary verb equivalent word to the noun equivalent word cut off from the auxiliary verb equivalent word by sentence division is performed. It will be possible.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a machine translation device including an embodiment of a document analysis device of the present invention.

FIG. 2 is a diagram for explaining a specific operation of the document analysis device shown in FIG.

FIG. 3 is a diagram for explaining a specific operation of the document analysis device shown in FIG.

FIG. 4 is a diagram for explaining a specific operation of the document analysis device shown in FIG.

5 is a diagram for explaining a specific operation of the document analysis device shown in FIG.

FIG. 6 is a diagram for explaining a specific operation of the document analysis device shown in FIG.

FIG. 7 is a diagram for explaining a specific operation of the document analysis device shown in FIG.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 input means 2 dictionary 3 generation means 4 output means 5 analysis means 6 analysis result display means 7 auxiliary verb sharing designation input means 51 analysis unit 52 auxiliary verb analysis unit

Claims (2)

[Claims] 1. An analysis unit in an analysis unit that performs morphological analysis, syntactic analysis, semantic analysis, and the like on an input sentence of a document processing device, and an analysis of auxiliary verb equivalent words during the analysis process by the analysis unit. Is detected in accordance with the processing rule based on the appearance pattern of the auxiliary verb equivalent word and the adjective equivalent word, and the scope level of the auxiliary verb scope noun candidate and the auxiliary verb scope noun candidate is detected, and based on the detection, the auxiliary verb scope When the output sentence of the document processing device is corrected and the auxiliary verb analysis unit in the analysis means for performing certification, the auxiliary verb scope related information indicating the auxiliary verb scope candidate and the scope level detected by the auxiliary verb analysis unit is displayed. The analysis result display means and the user's instruction for the auxiliary verb scope related information displayed by the analysis result display means. A document analysis device, comprising: an auxiliary verb sharing designation input means for changing the content of an auxiliary verb scope recognized by the auxiliary verb analysis unit based on a rule. 2. A scope level indicating two levels of information, "high" and "low" is handled, and the scope level is "high".
Is the auxiliary verb analysis unit that recognizes an auxiliary verb scope that recognizes an auxiliary verb scope candidate phrase as an auxiliary verb scope phrase, and whether the scope level of the auxiliary verb scope candidate phrase in the auxiliary verb scope misidentification sentence is "high". The document analysis apparatus according to claim 1, further comprising: the analysis result display unit that forms a display of auxiliary verb scope related information by a display that distinguishes whether it is "low" and a display of a candidate list.