JP2014067179A - Document processor and document processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To determine a place which is easily subjected to mistranslation in document data.SOLUTION: A document processor includes: input means for inputting document data; analysis means for analyzing a sentence of the document data input by the input means; affix extraction means for extracting a predetermined affix from the analysis result of the analysis means; rule storage means for storing an affix inspection rule in which a determination reference for determining whether or not a word including the predetermined affix is easily subjected to mistranslation is associated with the predetermined affix and a compound word extraction rule for extracting a compound word including an affix which is easily subjected to mistranslation; determination means for, when the word including the affix extracted from the analysis result satisfies the determination reference in the affix inspection rule, determining that the extracted affix is an affix which is easily subjected to mistranslation; and compound word extraction means for, when the word including the affix extracted from the analysis result satisfies the compound word inspection rule, extracting a compound word including the extracted affix following the rule as a compound word which is easily subjected to mistranslation.

Description

  Embodiments described herein relate generally to a document processing apparatus and a document processing program used for machine translation of document data.

  Conventionally, when machine translation of document data is performed, if compound words that are not registered in the translation dictionary include affixes that are difficult to translate directly, such as “Yes”, “No”, “No”, etc. It becomes difficult for the engine to determine the meaning of the compound word, and mistranslation is likely to occur.

  A compound word is a word in which two or more originally independent words are combined to have a new meaning and function as one word. Especially in technical sentences, terms that are compound words are often used. In general, since there are many types of compound words, it is difficult to manually register compound words in a dictionary manually.

  When a Japanese-English machine translation is performed on the first example of a compound word that is easy to mistranslate, “Document Management System Not Installed”, the translation result may be “Department introduced a document management system not”. is there. The cause of this mistranslation is that the machine translation engine does not correctly understand the abbreviation “not yet” of “not yet installed” in the “document management system not yet introduced section”.

  In addition, if the compound word “document management system unintroduced department” is translated by machine translation between Japan and China, the translation result may be mistranslated as “sentence management system sheep promotion department”. In this translation, the translation engine translates the abbreviation “not yet” of “not yet introduced” as “cotton sheep”.

Moreover, as a second example of a compound word that is easily mistranslated, if the “translation source pattern” is translated by Japanese-English machine translation, the type of machine translation engine to be used is different. Different results.
Translation A: the former pattern of conversion.
Translation B: the pattern of a changing agency.
The reason why the translation differs depending on the type of machine translation engine to be used is that the understanding of the machine translation engine for the affix “source” of “conversion source” in the “conversion source pattern” described above differs for each type. In the example shown in the translation A, the machine translation engine understands that the “conversion source pattern” is the “conversion source pattern”. On the other hand, in the example shown in the translated sentence B, the machine translation engine understands that the “conversion source pattern” is the “conversion source pattern”. Such fluctuations in understanding of the original text occur not only in machine translation but also in manual translation.

  To solve such mistranslation problems, if a compound word that is likely to be mistranslated can be automatically found in the source sentence before the compound word is processed by machine translation, this compound word is registered in the translation dictionary. By doing so, the accuracy of translation can be improved.

  For the diagnosis of compound words that are easily mistranslated, for example, a technique for extracting unregistered terms (including compound words) using dictionary information that frequently appears on the basis of the arrangement of parts of speech or character types is disclosed. ing.

In addition, a technique is disclosed that uses word co-occurrence information of a target language of translation to output a translation word candidate for an untranslated character string in the original language as a translation.
In addition, it detects compound words that are joined by interposing special characters such as hyphens from the original translation, draws a dictionary for each unregistered compound word, and translates it using the compound word structure information obtained from the result. A technique for outputting knowledge for use is disclosed.

JP 2002-342321 A JP 2000-259630 A Japanese Patent Application Laid-Open No. Hei 6-295311

  Since the technique for extracting unregistered terms described above does not consider affixes, all unregistered words can be extracted, and can be translated directly, that is, compound words that do not require registration in the dictionary (for example, “ Conversion pattern ") is also output. Therefore, when registering a compound word in the dictionary, it is necessary to manually remove a word that does not need to be registered, which is troublesome.

  In addition, the above-described technique for outputting candidate translations for untranslated character strings in the source language in the translated text only diagnoses untranslated words that are output in the source language in the translated text. As in the first and second examples of easy-to-use compound words, it is not possible to find a compound word that does not include an untranslated word in the translation result.

  Furthermore, the technology that outputs the knowledge for translation using the structure information of compound words mentioned above uses special characters such as hyphens as a clue for diagnosis, so it cannot find compound words with continuous kanji and kana in Japanese. .

  The problem to be solved by the present invention is to provide a document processing apparatus and a document processing program capable of determining a mistranslatable portion in document data.

  According to the embodiment, the document processing apparatus extracts a predetermined affix from an input unit for inputting document data, an analysis unit for analyzing a sentence of the document data input by the input unit, and an analysis result by the analysis unit. Affix extraction means, an affix inspection rule that associates a criterion for determining whether a word including the predetermined affix is easily mistranslated with the predetermined affix, and a compound for extracting a compound word including an affix that is easily mistranslated A rule storage means for storing a word extraction rule, and when a word including the affix extracted from the analysis result satisfies a criterion in the affix check rule, the extracted affix is determined to be an affix that is easily mistranslated. When the word including the affix extracted from the analysis means and the determination result satisfies the compound word inspection rule, the compound word including the extracted affix according to the rule is easily mistranslated. With a compound word extraction means for extracting as if language.

FIG. 2 is a block diagram illustrating an example of a hardware configuration of the document processing apparatus according to the first embodiment. FIG. 3 is a block diagram illustrating an example of a functional configuration of the document processing apparatus according to the first embodiment. The figure which shows an example of the affix dictionary stored in the affix dictionary storage part of the document processing apparatus in 1st Embodiment by a table format. The figure which shows an example of the affix check rule stored in the diagnostic rule storage part of the document processing apparatus in 1st Embodiment by a table format. The figure which shows an example of the compound word extraction rule stored in the diagnostic rule storage part of the document processing apparatus in 1st Embodiment by a table format. 6 is a flowchart illustrating an example of a processing operation procedure of the document processing apparatus according to the first embodiment. 6 is a diagram illustrating an example of a syntax analysis result of an input sentence by the document processing apparatus according to the first embodiment. FIG. The figure which shows an example of the diagnostic result of the compound word which is easy to be mistranslated by the document processing apparatus in 1st Embodiment. FIG. 9 is a block diagram illustrating an example of a functional configuration of a document processing apparatus according to a second embodiment. The figure which shows an example of the affix check rule stored in the diagnostic rule storage part of the document processing apparatus in 2nd Embodiment by a table format. The figure which shows an example of the compound word extraction rule stored in the diagnostic rule storage part of the document processing apparatus in 2nd Embodiment by a table format. 10 is a flowchart illustrating an example of a processing operation performed by the document processing apparatus according to the second embodiment. The figure which shows an example of the morphological analysis result of the input sentence by the document processing apparatus in 2nd Embodiment.

Hereinafter, embodiments will be described with reference to the drawings.
(First embodiment)
First, the first embodiment will be described.
FIG. 1 is a block diagram illustrating an example of a hardware configuration of a document processing apparatus according to the first embodiment.
As shown in FIG. 1, the document processing apparatus 30 in the first embodiment includes a computer 10 and an external storage device 20. The computer 10 is connected to the external storage device 20. The external storage device 20 stores a program (document processing program) 21 executed by the computer 10. The external storage device 20 is a hard disk drive or a non-volatile memory.
For example, the document processing apparatus 30 has a function of presenting a sentence designated by the user, receiving an instruction for diagnosing a compound word that is easily mistranslated, and outputting a diagnosis result.

  FIG. 2 is a block diagram illustrating a functional configuration example of the document processing apparatus according to the first embodiment. As shown in FIG. 2, the computer 10 includes an input unit 31, a syntax analysis unit 32, an affix extraction unit 33, an affix check unit 34, a compound word extraction unit 35, an output unit 36, an affix dictionary storage unit 37, and a diagnostic rule storage. Part 38 is included. In the present embodiment, these units are realized by the computer 10 executing the program 21 stored in the external storage device 20 shown in FIG.

  The program 21 can be stored in advance in a computer-readable storage medium and distributed. Further, the program 21 may be downloaded by the computer 10 via a network, for example. In the present embodiment, the affix dictionary storage unit 37 and the diagnostic rule storage unit 38 are stored in, for example, the external storage device 20 shown in FIG.

In the affix dictionary storage unit 37, an affix dictionary in which affixes and their types are registered is stored in advance. FIG. 3 is a diagram illustrating an example of the affix dictionary stored in the affix dictionary storage unit 37 of the document processing apparatus according to the first embodiment in a table format.
As shown in FIG. 3, the affix dictionary describes a plurality of types of headings and affix types corresponding to the headings. A headline described in an affix dictionary is an affix that is easily mistranslated when in a predetermined dependency with another word. The affix type represents the type of affix and is information used when referring to an affix inspection rule described later. In the example of FIG. 3, there are three types of affix types “A”, “B”, and “C”, the affix headings “T” and “Non” belong to the affix type “A”, and the affix headings “unknown” ”,“ No ”,“ Time ”, and“ Previous ”belong to the affix type“ B ”, and the affix headings“ Yes ”and“ Original ”belong to the affix type“ C ”.

In the diagnostic rule storage unit 38, affix checking rules and compound word extraction rules are stored in advance.
The affix checking rule is a rule for determining whether the affix to be checked in the input sentence is easily mistranslated.
FIG. 4 is a diagram showing an example of the affix check rule stored in the diagnosis rule storage unit 38 of the document processing apparatus according to the first embodiment in a table format.
As shown in FIG. 4, in the affix check rule, for each affix type described in the affix dictionary, a criterion for determining whether or not the affix to be inspected corresponding to this affix type is likely to be mistranslated. Prepared. In the example illustrated in FIG. 4, when the affix type is “A”, “if the affix node and the parent node are“ collocation ”, it is determined that mistranslation is likely to occur. Is a determination criterion.

  In addition, the affix dictionary as shown in FIG. 3 and the affix check rule as shown in FIG. Also good. Further, as described above, it is a matter of course that the configuration of the affix check rule can be simplified by determining the affix type and using the affix dictionary and the affix check rule, respectively.

FIG. 5 is a diagram illustrating an example of a compound word extraction rule stored in the diagnosis rule storage unit 38 of the document processing apparatus according to the first embodiment in a table format.
The compound word extraction rule shown in FIG. 5 is a rule for extracting a compound word by determining a boundary of a compound word including an affix that is easily mistranslated.

The input unit 31 receives input of document data in response to a user operation on, for example, a keyboard or a mouse.
The syntax analysis unit 32 performs syntax analysis on the input sentence in the document data input by the input unit 31, and outputs an analysis result.

  The affix extraction unit 33 extracts an affix based on the affix dictionary stored in the affix dictionary storage unit 37 when an affix to be examined is included in the input sentence. The affix is subject to an affix check.

  The affix checking unit 34 is stored in the diagnostic rule storage unit 38 by using the affix type of the affix to be subjected to the affix check extracted by the affix extraction unit 33 with respect to the result of the syntax analysis by the syntax analysis unit 32. Refers to the affix checking rule. The affix checking unit 34, when the word including the affix to be inspected, which is indicated by the syntax analysis result by the syntax analysis unit 32, matches the criterion corresponding to the affix type of the affix to be inspected in the affix checking rule, It is assumed that the corresponding affix in the applicable word is easy to mistranslate.

Based on the compound word extraction rules stored in the diagnostic rule storage unit 38, the compound word extraction unit 35 determines the boundaries of compound words including affixes that are easily mistranslated, and extracts compound words that are easily mistranslated.
The output unit 36 uses a compound word diagnosis result, that is, an affix that is easily mistranslated, which is a test result by the affix checking unit 34, and a compound word that is easily mistranslated extracted by the compound word extraction unit 35. Output for users.

  Next, referring to the flowchart shown in FIG. 6, a document according to the present embodiment when a compound word diagnosis is performed based on a syntax analysis result for a sentence input by the user, a diagnosis result is generated, and the result is output to the user. A processing procedure of the processing apparatus will be described.

(1) Acquisition of Input Sentence When the user performs an operation on the input unit 31 to input a sentence to be examined, the input unit 31 acquires the input sentence (step S1). This sentence may be input directly by the user from a keyboard or may be read from an existing file.

(2) Syntax analysis The syntax analysis unit 32 performs syntax analysis on the input sentence (step S2).
FIG. 7 is a diagram illustrating an example of a syntax analysis result when the input sentence by the document processing apparatus according to the first embodiment is “output the conversion source pattern before registration”.
In the ellipse in the parsing result shown in FIG. 7, the stem of each clause of the input sentence is written. This is called a node. In FIG. 7, two nodes having a dependency relationship are connected by an arrow. This arrow is called an arc. A node connected to the end of the arrow is called a parent node, and a node connected to the arrow is called a child node. The word enclosed in angle brackets (<>) in the ellipse is the part of speech of the node corresponding to this ellipse.

  Further, an explanation of the dependency relationship between two nodes connected by this arrow is given to the arrow between the nodes. For example, as shown in FIG. 7, when “wo” is attached to the arrow between the child node “pattern” and the parent node “output”, the child node “pattern” connected by the arrow is This means that the dependency relationship with the parent node “output” is unequaled.

  Also, as shown in FIG. 7, when “collocation” is attached to the arrow between the child node “noun” and the parent node “previous”, the child node “registration” and the parent node connected by this arrow “Previous” means a collocation relationship. Similarly, the child node “transformation” with “cold” attached to the arrow and the parent node “original” have a collocation relationship, and the child node “original” with “cold” attached to the arrow This means that the parent node “pattern” has a collocation relationship. Here, the collocation relationship refers to a linguistic expression that is composed of a plurality of morphemes but is used in the same way as words.

(3) Affix extraction Whether the affix extraction unit 33 includes the affix registered in the affix dictionary in the syntax analysis result of the input sentence while referring to the headings of the affix dictionary stored in the affix dictionary storage unit 37 Is determined (step S3). When the affix registered in the affix dictionary is included in the parsing result, the affix extraction unit 33 sets the corresponding affix as an inspection target.

  For example, the syntax analysis result of the input sentence “output the conversion source pattern before registration” described above includes the affix “original” and the affix “previous” registered in the affix dictionary. . When referring to the affix dictionary shown in FIG. 3, the affix extraction unit 33 includes both the affix “original” and the affix “previous” included in the input sentence “output the conversion source pattern before registration”. Both are subject to affix inspection.

(4) Affix check When the input sentence contains an affix to be checked, the affix check unit 34 determines the affix check rule stored in the diagnostic rule storage unit 38 based on the syntax analysis result of the input sentence. Apply and check whether the syntax analysis result matches the criterion corresponding to the affix type of the affix to be inspected in this affix inspection rule (step S4). When the parsing result matches the criterion, the affix checking unit 34 determines that the affixed as the subject of inspection is an affix that is easily mistranslated.

  For example, the affix type of the affix “original” that is the object of inspection in the input sentence “output the conversion source pattern before registration” is “C” in the affix dictionary shown in FIG. The unit 34 collates the determination criterion corresponding to the affix type “C” of the affix check rule illustrated in FIG. 4 with the syntax analysis result. In the example shown in FIG. 4, the determination criterion corresponding to the affix type “C” is “if the affix node and its child node have a“ collocation ”dependency relationship, it is determined that mistranslation is likely to occur. Is.

  As shown in FIG. 7, the “original” node, which is the affix to be inspected in the syntax analysis result, has a dependency relationship with the child node “transformation” when this node is the parent node. Therefore, the affix checking unit 34 determines that “original”, which is the affix to be inspected in the syntax analysis result, is an affix that is easily mistranslated.

  On the other hand, since the affix type of the affix “previous” to be examined in the input sentence “output the conversion source pattern before registration” is “B” in the affix dictionary shown in FIG. 34 collates the determination standard corresponding to the affix type “B” of the affix check rule shown in FIG. 4 with the syntax analysis result. In the example illustrated in FIG. 4, the determination criterion for the affix type “B” is that it is easy to make a mistranslation if the parent node and the child node of the affix node are “collocation”. Is. Specifically, this criterion indicates that if an affix node and its parent node are a “collocation” dependency and this affix node and its child node are also a “collocation” dependency, it is determined that it is easy to mistranslate. .

  As shown in FIG. 7, the “previous” node, which is the affix to be inspected in the syntax analysis result, has a dependency relationship with the child node “registration” when this node is the parent node. However, the “previous” node, which is an affix in the parsing result, does not have a dependency relationship with the parent node “transformation” when this node is a child node. Therefore, the affix checking unit 34 does not determine that the affix “previous”, which is the affix to be inspected in the syntax analysis result, is easily mistranslated.

(5) Compound Word Extraction Next, the compound word extraction unit 35 stores diagnostic rules based on the syntax analysis result of the input sentence when an affix that is easily mistranslated is included in the input sentence as a result of processing by the affix checking unit 34. The compound word extraction rule stored in the unit 38 is used to determine the boundary of the compound word including the corresponding affix, and the compound word is extracted from the input sentence (step S5).

  For example, the compound word extraction unit 35 includes the affix “original” that is easily mistranslated as described above based on the compound word extraction rule shown in FIG. 5 with respect to the syntax analysis result shown in FIG. Determine word boundaries. The compound word extraction rule shown in FIG. 5 is “From affix, nodes in collocation are grouped together into a single compound word”. In the syntax analysis result shown in FIG. 7, the compound node is extracted because the child node having the collocation relationship from the affix “original” is “transformation” and the parent node having the collocation relationship from the affix “form” is “pattern”. According to the compound word extraction rule, the unit 35 sets the “conversion source pattern” of the input sentence to a compound word including the affix “original”. Since this compound word includes an affix that is easily mistranslated, the compound word is easily mistranslated.

(6) Output The output unit 36 outputs a diagnostic result including an affix that is easily mistranslated by the affix checking unit 34 and a compound word extracted by the compound word extracting unit 35 (step S6). The diagnosis result is output by, for example, displaying on a liquid crystal display, outputting a list of diagnosis results as a CSV file, or writing an indication message as a comment of a document file.

FIG. 8 is a diagram illustrating an example of a compound word diagnosis result that is easily mistranslated by the document processing apparatus according to the first embodiment.
As shown in FIG. 8, in the diagnosis result, there is a message indicating that the input sentence includes a compound word that is not registered in the translation dictionary, and that this compound word includes an affix that is easily mistranslated. It is shown that this compound word is a compound word that is easily mistranslated.

  As described above, in the document processing apparatus according to the first embodiment, the user can find an affix that is easily mistranslated and a compound word that is easily mistranslated based on the syntax analysis result of the input sentence. Then, the accuracy of subsequent machine translation can be improved by registering a compound word that is easily mistranslated in the translation dictionary based on the diagnosis result. As described above, in the document processing apparatus according to the first embodiment, the terms to be registered in the translation dictionary can be automatically extracted, so that the burden on the user for expanding the translation dictionary can be reduced. Furthermore, since the user can find difficult-to-understand compound words from the input sentence, it can support the improvement of the document quality.

(Second Embodiment)
Next, a second embodiment will be described. Note that, in the configuration of the document processing apparatus according to this embodiment, the description of the same parts as those of the first embodiment is omitted.
FIG. 9 is a block diagram illustrating a functional configuration example of the document processing apparatus according to the second embodiment. As shown in FIG. 9, the computer 10 of the document processing apparatus 30 according to the second embodiment includes an input unit 311, a morpheme analysis unit 312, an affix extraction unit 313, an affix check unit 314, a compound word extraction unit 315, and an output unit 316. , An affix dictionary storage unit 317, and a diagnostic rule storage unit 318.

  Compared with the first embodiment, the document processing apparatus according to the second embodiment includes a morpheme analysis unit 312 instead of the syntax analysis unit 32. In this embodiment, the affix dictionary storage unit 317 and the diagnostic rule storage unit 318 are stored in, for example, the external storage device 20 illustrated in FIG. The affix dictionary storage unit 317 stores an affix dictionary as shown in FIG. 3 as in the first embodiment.

  In this embodiment, the affix check rule and the compound word extraction rule stored in the diagnostic rule storage unit 318 are rules based on the result of morphological analysis, and are based on the syntax analysis result described in the first embodiment. It is different from affix checking rules and compound word extraction rules.

FIG. 10 is a diagram illustrating an example of the affix check rule stored in the diagnosis rule storage unit of the document processing apparatus according to the second embodiment in a table format.
In the affix inspection rule shown in FIG. 10, for each affix type described in the affix dictionary, a criterion for determining whether the affix to be inspected corresponding to this affix type is likely to be mistranslated is prepared. The In the example illustrated in FIG. 10, when the affix type is “A”, “if the part of speech of the morpheme immediately after the affix is“ noun ”, it is determined that it is easy to mistranslate. There is a criterion called “

FIG. 11 is a diagram illustrating an example of a compound word extraction rule stored in the diagnosis rule storage unit of the document processing apparatus according to the second embodiment in a table format.
The compound word extraction rule shown in FIG. 11 is a rule for determining a boundary of a compound word including an affix that is easily mistranslated and extracting the compound word.

Similar to the input unit 31, the input unit 311 accepts an instruction from the user, for example, according to a user operation on a keyboard or a mouse.
The morpheme analysis unit 312 performs morpheme analysis on the input sentence and outputs an analysis result.
Based on the affix dictionary stored in the affix dictionary storage unit 317, the affix extraction unit 313 extracts an affix to be examined when it is included in the input sentence. The affix is subject to an affix check.

The affix checking unit 314 uses the affix type of the affix to be examined included in the input sentence extracted by the affix extraction unit 313 with respect to the result of the morpheme analysis by the morpheme analysis unit 312, and the diagnostic rule storage unit The affix check rule stored in 318 is referred to. When the affix checking unit 314 includes the affix to be inspected, which is indicated by the result of the morpheme analysis by the morpheme analysis unit 312, applies to the criterion corresponding to the affix type of the affix to be inspected in the affix checking rule Suppose that the corresponding affix in this applicable word is easy to mistranslate.
Based on the compound word extraction rules stored in the diagnostic rule storage unit 318, the compound word extraction unit 315 determines the boundaries of compound words including affixes that are easily mistranslated, and extracts compound words that are easily mistranslated.

  The output unit 316 uses the compound word diagnosis result, that is, the affix that is easily mistranslated, which is the test result by the affix checking unit 314, and the compound word that is easily mistranslated extracted by the compound word extraction unit 315 for the user. Output for.

FIG. 12 is a flowchart illustrating an example of a processing operation performed by the document processing apparatus according to the second embodiment.
(1) Acquisition of Input Sentence First, when a sentence to be examined is input by a user, the input unit 311 acquires the input sentence (step S1). This sentence may be input directly by the user from a keyboard or may be read from an existing file.

(2) Morphological Analysis Next, the morphological analysis unit 312 performs morphological analysis on the input sentence (step S12). Here, it is assumed that the input sentence is “output the conversion source pattern before registration” described in the first embodiment.

FIG. 13 is a diagram illustrating an example of a morphological analysis result when the input sentence by the document processing apparatus according to the second embodiment is “output the conversion source pattern before registration”.
As shown in FIG. 13, in the morpheme analysis result, the input sentence is divided in units of morpheme by a diagonal line “/”. In the morphological analysis results, words enclosed in angle brackets (<>) are morpheme parts of speech.

(3) Affix extraction The affix extraction unit 313 refers to the heading of the affix dictionary stored in the affix dictionary storage unit 317 based on the morphological analysis result of the input sentence, as in the first embodiment. It is determined whether the affix registered in the affix dictionary is included in the morphological analysis result of the input sentence. If the affix registered in the affix dictionary is included in the morpheme analysis result, this affix is the affix to be checked. (Step S13).

  For example, it is understood that the affix “original” and the affix “previous” registered in the affix dictionary are included in the morphological analysis result of the input sentence “output the conversion source pattern before registration”. . When the affix extraction unit 313 refers to the affix dictionary shown in FIG. 3, both the affix “original” and the affix “previous” included in the input sentence “output the conversion source pattern before registration” are included in the affix dictionary. Both are subject to affix inspection.

(4) Affix check When the input sentence includes an affix to be checked, the affix check unit 314 determines the affix check rule stored in the diagnostic rule storage unit 318 based on the morphological analysis result of the input sentence. It is applied to check whether the morphological analysis result applies to the determination criterion corresponding to the affix type of the affix to be inspected in this affix inspection rule (step S14). The affix checking unit 314 determines that the affix is an affix that is likely to be mistranslated when the morphological analysis result meets the criteria.

  For example, the affix type of the affix “original” that is the object of inspection in the input sentence “output the conversion source pattern before registration” is “C” in the affix dictionary shown in FIG. The unit 314 collates the determination standard corresponding to the affix type “C” of the affix inspection rule illustrated in FIG. 10 with the morphological analysis result.

  In the example illustrated in FIG. 10, the determination criterion corresponding to the affix type “C” is “if the part of speech of the morpheme immediately before the affix is“ noun ”, it is determined that mistranslation is likely to occur. Is. In the morpheme analysis result shown in FIG. 13, since the part of speech of the morpheme “transformation” immediately before the affix “original” to be inspected is “noun”, the affix inspection unit 314 checks the affix “original” to be inspected in the morpheme analysis result. Is an affix that is easy to mistranslate.

  On the other hand, since the affix type of the affix “previous” to be examined in the input sentence “output the conversion source pattern before registration” is “B” in the affix dictionary shown in FIG. In step 314, the determination criterion corresponding to the affix type “B” in the affix inspection rule illustrated in FIG. 10 is collated with the morphological analysis result.

  In the example shown in FIG. 10, the criterion for the affix type “B” is “if the morpheme immediately before and after the affix both have“ nouns ”, it is determined that they are easily mistranslated. Is. As shown in FIG. 13, the part of speech of the morpheme “registration” immediately before the affix “previous” which is the affix to be examined in the morpheme analysis result is “noun”, but the morpheme “no” immediately after the affix “previous” The part of speech is not a "noun". Accordingly, the affix checking unit 314 does not set the affix “previous” to be checked in the morphological analysis result as an affix that is easily mistranslated.

(5) Compound word extraction The compound word extraction unit 315 stores the diagnosis rule storage unit 318 based on the morphological analysis result of the input sentence when an affix that is easily mistranslated is included in the input sentence as a result of processing by the affix checking unit 314. Using the stored compound word extraction rule, the boundary of the compound word including the corresponding affix that is easily mistranslated is determined, and the compound word is extracted (step S15).

  For example, the compound word extraction unit 315 determines the boundary of the compound word including the affix “original” based on the compound word extraction rule shown in FIG. 11 with respect to the morphological analysis result shown in FIG. Since the compound word extraction rule shown in FIG. 11 is “a morpheme whose part of speech before and after the affix is a noun is combined into one compound word”, the compound word extraction unit 315 uses this rule to The “conversion source pattern” is extracted as a compound word including the affix “original” that is easily mistranslated. Since this compound word includes an affix that is easily mistranslated, the compound word is easily mistranslated.

(6) Output As in the first embodiment, the output unit 316 includes an affix that is easy to mistranslate discovered by the affix checking unit 314 and a compound word extracted by the compound word extraction unit 315. The result is output as shown in FIG. 8 (step S16).

  As described above, the document processing apparatus according to the second embodiment is likely to be mistranslated in the input sentence using only the morphological analysis result of the input sentence without using the parsing as in the first embodiment. Discover compound words. In addition, compared to syntax analysis, morphological analysis can reduce the cost for realization, and therefore, in the second embodiment, the cost for realization is lower than that in the first embodiment. There are advantages.

  The first embodiment and the second embodiment use syntax analysis and morphological analysis, respectively, but the analysis method is not limited to these. For example, any analysis that can identify information necessary to extract a compound word that is not registered in the dictionary and that includes a careful affix may be used.

According to each of these embodiments, it is possible to provide a document processing apparatus and a document processing program capable of determining a portion that is easily mistranslated in document data.
The storage medium for distributing the program may be in any form as long as the storage medium can store the program and can be read by the computer.

  In addition, an OS (operating system) running on a computer based on an instruction of a program installed in the computer from a storage medium, MW (middleware) such as database management software, network software, and the like realize the above-described embodiment. A part of each process may be executed.

  Furthermore, the storage medium is not limited to a medium independent of the computer, but also includes a storage medium in which a program transmitted via a LAN or the Internet is downloaded and stored or temporarily stored.

  Further, the number of storage media is not limited to one, and the case where the processing in the above embodiment is executed from a plurality of media is also included in the storage media in the present invention, and the media configuration may be any configuration.

  The computer executes each process in the above embodiment based on a program stored in a storage medium, and includes any one device such as a personal computer or a system in which a plurality of devices are connected to a network. It may be configured as follows.

  Further, the computer is not limited to a personal computer, but includes an arithmetic processing device, a microcomputer, and the like included in an information processing device, and is a general term for devices and devices that can realize the functions of the present invention by a program.

  Although several embodiments of the invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  DESCRIPTION OF SYMBOLS 10 ... Computer, 20 ... External storage device, 21 ... Program, 30 ... Document processing device, 31, 311 ... Input part, 32 ... Syntax analysis part, 33, 313 ... Affix extraction part, 34, 314 ... Affix check part, 35 , 315 ... compound word extraction unit, 36, 316 ... output unit, 37, 317 ... affix dictionary storage unit, 38, 318 ... diagnostic rule storage unit, 312 ... morpheme analysis unit.

Claims (5)

An input means for inputting document data;
Analyzing means for analyzing a sentence of the document data input by the input means;
Affix extraction means for extracting a predetermined affix from the analysis result by the analysis means;
An affix checking rule that associates a criterion for determining whether a word including the predetermined affix is easily mistranslated with the predetermined affix, and a compound word extraction rule for extracting a compound word including an affix that is easily mistranslated Rules storage means to
A determination unit that determines that the extracted affix is an affix that is likely to be mistranslated when a word including the affix extracted from the analysis result satisfies a determination criterion in the affix check rule;
When a word including an affix extracted from the analysis result satisfies the compound word inspection rule, a compound word extracting unit extracts a compound word including the extracted affix as a compound word that is easily mistranslated according to the rule. And a document processing apparatus. Affix dictionary storage means for storing an affix dictionary in which a heading of a predetermined affix and a type of affix of the heading are associated;
The affix extraction means includes
2. The document according to claim 1, wherein when a word included in an analysis result by the analyzing unit matches a heading of a predetermined affix in the affix dictionary, the matching word is extracted as the predetermined affix. Processing equipment. The analysis means includes
Parsing the sentence of the document data input by the input means,
The determination means includes
When a word including the extracted affix from the result of the parsing satisfies the determination criteria in the affix checking rule, the extracted affix is determined to be an affix that is easily mistranslated,
The compound word extraction means includes
When a word including the extracted affix satisfies the compound word check rule from the result of the parsing, the compound word including the extracted affix according to the rule is extracted as a compound word that is easily mistranslated. The document processing apparatus according to claim 1. The analysis means includes
Perform morphological analysis of the sentence of the document data input by the input means,
The determination means includes
When a word including the extracted affix from the result of the morphological analysis satisfies a determination criterion in the affix check rule, the extracted affix is determined to be an affix that is easily mistranslated,
The compound word extraction means includes
When a word including the extracted affix satisfies the compound word check rule from the result of the morphological analysis, a compound word including the extracted affix according to the rule is extracted as a compound word that is easily mistranslated. The document processing apparatus according to claim 1. Computer
Input means for inputting document data,
Analyzing means for analyzing a sentence of the document data input by the input means;
Affix extraction means for extracting a predetermined affix from the analysis result by the analysis means,
An affix checking rule that associates a criterion for determining whether a word including the predetermined affix is easily mistranslated with the predetermined affix, and a compound word extraction rule for extracting a compound word including an affix that is easily mistranslated A rule storage means,
When a word including an affix extracted from the analysis result satisfies a determination criterion in the affix check rule, a determination unit that determines that the extracted affix is an affix that is easily mistranslated, and extracted from the analysis result Document processing for functioning as a compound word extraction means for extracting a compound word including the extracted affix as a compound word that is easily mistranslated according to the rule when a word including an affix satisfies the compound word inspection rule program.

Images