JPH07200592A - Text processor - Google Patents

Abstract

PURPOSE:To provide a text processor capable of presuming the function of a word in a sentence and correcting the word to a correct notation for the word for which certain expression is recommended. CONSTITUTION:A HIRAGANA (the cursive form of Japanese syllabary) writing recommended word retrieval, part 2 uses a HIRAGANA writing recommended word dictionary 3 and retrieves a HIRAGANA writing recommended word from the morpheme information of the sentence stored in a morepheme information storage part 1. Then, an erroneous use detection part 4 detects the erroneous use of the notation by using an erroneous use detection rule 5, syntax information inside a syntax information storage device 7 and the information of mandatory cases by a mandatory case retrieval part 8, etc., for the HIRAGANA writing recommended word retrieved in the HIRAGANA writing recommended word retrieval part 2. In this case, the mandatory case retrieval part 8 uses the syntax information inside the syntax information storage part 7, a mandatory case dictionary 10 and a semantic kind dictionary 9 and investigates the presence/absence of the mandatory case. When the error of the notation is detected in the erroneous use detection part 4, that effect is outputted to an output part 6 and rewrite is performed in a character kind rewrite part 11 corresponding to the instruction of the rewrite from an input part 12 by a user.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a sentence processing apparatus capable of detecting a word or the like which is recommended for a certain expression format and converting it into a correct notation when creating and reviewing a sentence.

[0002]

2. Description of the Related Art In Japanese, there are cases where a certain expression format is recommended when writing sentences. For example, there are recommended words for writing in hiragana that are said to be preferable to write in hiragana. As a specific example, conjunctions such as "but" and "if" are
Writing in hiragana is easier to understand than writing “side by side” or “young” in kanji. Therefore, Hiragana writing is recommended. However, some hiragana writing recommended words, such as "hence," recommend hiragana writing only when acting as a conjunction. Of course, when working as a verb, it is easier to understand in "kanji" in Kanji to distinguish it from conjunctions.

[0003] Conventionally, as a technique related to such character type conversion, for example, as described in Japanese Patent Laid-Open No. 1-207868, a Kanji that is not a commonly used Kanji is corrected to Hiragana, or Japanese Patent Laid-Open No. 5-35729. As described in Japanese Patent Publication, there is a device that detects a kana writing word in which kanji or katakana notation is desirable and automatically corrects and displays it.

However, none of the above-mentioned conventional techniques considers the action of the word to be corrected in the sentence, so that a desirable notation such as the above-mentioned recommended word for hiragana writing depends on the role in the sentence. However, there is a drawback in that words that change with time cannot always be corrected to the desired notation.

[0005]

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned circumstances. For a word for which a certain expression is recommended, the action in the text of the word is guessed and the word is corrected to the correct notation. It is an object of the present invention to provide a text processing device that can be used.

[0006]

According to the present invention, information about a morpheme of a sentence is input, and the first expression recommended word in which the first expression in the sentence is recommended is replaced with a part described in another expression format. In a sentence processing device for detection, a first expression recommended word dictionary in which a description of a first expression of a first expression recommended word and a description of another expression format are described, and the first expression recommended word dictionary is used for input. First expression recommended word searching means for searching the first expression recommended word from the information on the morpheme, and the expression form of the first expression recommended word from the position of the first expression recommended word and the part of speech of words before and after Misuse detection rule holding means for holding the misuse detection rule in which knowledge whether it should be
It is characterized by comprising misuse detecting means for detecting misuse of the first expression recommended word using the information on the morpheme and the misuse detection rule.

[0007] Further, information about the syntax of the sentence is input, and the misuse detection rule determines whether the first expression recommended word may act as a word, in addition to the position of the first expression recommended word and the part of speech of words before and after. Describes the knowledge as to which expression format the first expression recommended word should be described,
The misuse detection means may be configured to detect misuse of the first recommended expression word by using information about a morpheme and syntax of an input sentence and the misuse detection rule.

Further, an essential case dictionary in which a case required when the first expression recommended word or the like is used as a noun is described, and a meaning type dictionary in which a meaning type to which the word belongs is described.
When the first expression recommended word searched by the first expression recommended word searching means may act as a verb, the sentence using the information about the syntax of the inputted sentence, the essential case dictionary, and the meaning type dictionary An essential case search means for checking the existence of a word or phrase corresponding to the essential case of the idiom inside, wherein the misuse detection rule includes the position of the first expression recommended word, the part of speech of words before and after, and the first word.
In addition to the possibility that the recommended word of expression may act as a idiom,
From the presence or absence of the indispensable case when the recommended word for expression works as a word,
The knowledge as to which expression format the first expression recommended word should be described in is described, and the misuse detection means includes information about the morpheme of the input sentence, information about the syntax, and the essential case search means. It is also possible to detect misuse of the first expression recommended word by using the search result and the misuse detection rule.

[0009]

According to the present invention, the first expression recommended word search means searches the first expression recommended word using the first expression recommended word dictionary on the basis of the information about the morpheme of the input sentence and misuses it. The detecting means checks whether or not the notation of the first expression recommended word is incorrect. When a misused portion is detected in this way, it is possible to inform the user of that fact and convert the expression format according to the rewriting instruction from the user.

In the misuse detecting means, the information about the morpheme of the input sentence and the misuse detection rule are used as the information for checking whether or not the notation is incorrect. In addition to this, it is also possible to use information about the syntax, or information about the syntax of the sentence, and information such as the result of the search performed by the essential case search means using the essential case dictionary and the meaning type dictionary. As the misuse detection rule, knowledge about what expression format the first expression recommended word should be described is described, and as the information used for the determination, the position of the first expression recommended word and the part of speech of words before and after it. In addition, information about whether or not the first expression recommended word may work as a noun and whether or not the first expression recommended word has an indispensable case as a noun is described.

The first expression can be written in hiragana, for example, and can be configured to detect misuse between hiragana and kanji notation.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a block diagram showing an embodiment of a text processing apparatus of the present invention. In the figure, 1 is a morpheme information storage unit,
2 is a recommended hiragana writing word search unit, 3 is a recommended hiragana writing word dictionary, 4 is an misuse detection unit, 5 is an misuse detection rule, 6 is an output unit, 7 is a syntax information storage unit, 8 is an essential case search unit, and 9 is A semantic type dictionary, 10 is an essential case dictionary, 11 is a character type rewriting unit, and 12 is an input unit. In this embodiment, as an example, misuse between the Hiragana notation and the Kanji notation and rewriting thereof will be described.

The morpheme information storage unit 1 and the syntax information storage unit 7 store the sentences to be processed in a morpheme analyzed state and a syntactic analyzed state, respectively. The hiragana writing recommended word search unit 2 searches the morpheme information of the sentence to be processed stored in the morpheme information storage unit 1 using the hiragana writing recommended word dictionary 3 to search for hiragana writing recommended words. Hiragana writing recommended word dictionary 3
Stores hiragana writing and kanji writing in pairs for recommended hiragana writing words. The misuse detection unit 4 uses the misuse detection rule 5, the syntax information storage unit 7, and the indispensable case search unit 8 to write the recommended hiragana writing word for the hiragana writing recommended word searched by the hiragana writing recommended word search unit 2. Check for mistakes. The misuse detection rule 5 describes the knowledge as to whether or not the recommended hiragana writing word should be written in hiragana or kanji. As a rule for the judgment, the position of the recommended hiragana writing word in the sentence or Information such as the part of speech of the preceding and following words, the possibility that the recommended hiragana writing word works as a noun, and the presence or absence of the essential case when the recommended hiragana writing word works as a noun are described. The essential case search unit 8 uses the syntactic information stored in the syntactic information storage unit 7, the meaning type dictionary 9, and the essential case dictionary 10 to identify the recommended hiragana writing word searched by the hiragana writing recommended word search unit 2. If it is, it is checked whether there is a word corresponding to the essential case. In the semantic type dictionary 9, for example, a semantic type that is a category to which a semantic word belongs, such as whether a noun word represents a person, an animal, or a place name, is described. The indispensable case dictionary 10 describes the case that is required when a recommended word such as hiragana writing is used as a noun. The output unit 6 presents to the user the recommended hiragana writing words detected by the misuse detection unit 4 as misuse. The input unit 12 receives an input from the user, and notifies the character type rewriting unit 11 of, for example, an instruction input of rewriting the recommended hiragana writing word detected as misuse. The character type rewriting unit 11 is detected as misuse by the misuse detection unit 4, and when a rewriting instruction is given from the input unit 12, the morpheme information storage unit 1 and the syntax information storage unit 7 are executed.
Rewrite the recommended hiragana words in the sentences stored in.

In the above-mentioned structure, when the misuse detection unit 4 does not check the essential case when the recommended hiragana writing word works as a noun, the essential case search unit 8, the meaning type dictionary 9, and the essential case dictionary are used. It can be configured by omitting 10. Further, when the misuse detection unit 4 does not check the possibility that the recommended hiragana writing word works as a noun, the syntax information storage unit 7 can be omitted. In a configuration in which some of these are omitted, some of the information stored as the misuse detection rule 5 can also be omitted.

The morpheme information storage unit 1 can be composed of a sentence storage unit and a morpheme analysis unit for performing a morpheme analysis on a sentence. Similarly, the syntactic information storage unit 7 can also be configured by a syntactic analysis unit that performs syntactic analysis on the sentence stored in the sentence storage unit. Furthermore, through networks and lines,
It may be configured to be input from another computer or system. Also, it receives morpheme information and syntactic information of a sentence from a higher-level system such as a document input editing system,
It may be configured as a subsystem that passes the output of the misuse detection unit 4 to a higher-level system. In this case, the functions of the character type rewriting unit 11, the output unit 6, and the input unit 12 may be incorporated in a higher-level system.

FIG. 2 is a flow chart for explaining the outline of the operation in one embodiment of the text processing apparatus of the present invention. When an instruction to check the notation is given to the sentence processing device, first, in S21, the hiragana writing recommended word search unit 2 uses the hiragana writing recommended word dictionary 3 and the sentences stored in the morpheme information storage unit 1 are stored. Hiragana writing recommended words are searched from the morpheme information of.

Next, in S22, the misuse detection unit 4
Regarding the recommended hiragana writing words searched by the hiragana writing recommended word search unit 2, the notation is incorrect using the misuse detection rule 5, the syntax information in the syntax information storage device 7, the essential case information by the essential case search unit 8, and the like. Check for not. Here, the indispensable case search unit 8 checks whether or not the recommended hiragana writing word searched by the hiragana writing recommended word search unit 2 can be a noun, based on the syntax information stored in the syntax information storage unit 7. When it can be used as a dictation, the case required by the verb is searched from the essential case dictionary 10, and it is necessary to refer to the words around the recommended hiragana writing word from the syntax information stored in the syntax information storage unit 7. Check whether there is a case. If there is an essential case, it is checked using the meaning type dictionary 9 whether the essential case is semantically appropriate.

When an error is detected in the notation of the recommended hiragana writing word in S22, the fact is output to the output unit 6 to notify the user in S23. Then, the input unit 12
When a rewriting instruction is input from, in S24,
The character type rewriting unit 11 rewrites the recommended hiragana writing words into correct words. Then, the process returns to S21 in order to check the next recommended hiragana writing word. The process of inspection and rewriting ends when the sentence is completed.

The operation of the text processing apparatus of the present invention will be described based on a specific example. First, a case in which the first example sentence "Thus, this answer is true." Is input will be described with reference to FIGS. Here, a case will be described in which the word “follow” is rewritten in hiragana notation. Also, for this example sentence, the syntax information storage unit 7, the essential case search unit 8, the meaning type dictionary 9, the essential case dictionary 1
The case where the configuration does not use 0 will be described.

FIG. 3 is an explanatory diagram of the morphological analysis result of the first example sentence. The first example sentence is stored in the morpheme information storage unit 1 as morpheme information as shown in FIG. Further, this morpheme information may be provided from the beginning as shown in FIG. 3, or the first example sentence to be analyzed may be morpheme-analyzed and input. The stored morpheme information includes a surface layer indicating the expression itself used in the sentence, its part of speech, a stem, and its inflection form in the case of a verb. Other information can also be stored. In FIG. 3, a remarks column is provided. In the first example sentence, the word “according to” has the possibility of either being a conjunction or a verb + particle. Therefore, in FIG. 3, as the morpheme information of the word “according to”, information on both a conjunction and a verb + particle is stored.

FIG. 4 is an explanatory diagram of an example of the hiragana writing recommended word dictionary 3. The hiragana writing recommended word dictionary 3 is shown in FIG.
As shown in, the information in which the hiragana writing and the kanji writing of the recommended hiragana writing word are paired is stored. The hiragana writing recommended word search unit 2 performs pattern matching between the information of each surface layer of the morpheme information shown in FIG. 3 and the contents of the hiragana writing recommended word dictionary 3 shown in FIG. Here, since "according to" and "aru" are searched, these two words are used as recommended words for writing in hiragana, and these two words are checked for misuse in the next step.

FIG. 5 is an explanatory diagram of an example of the misuse detection rule 5. The misuse detection rule 5 includes, for example, as shown in FIG. 5, a headword, a rule when the headword is misused, and a rewrite word. Pre. Used in the description of rules in the figure. Part of speech, Post. The part-of-speech represents the part-of-speech of the word immediately before the headword and the part-of-speech immediately after it. Similarly, Pre. Surface or Pre. The inflectional form also indicates the surface layer immediately before the headword, the inflectional form. Also, φ indicates that it does not exist, and “T” indicates that it is always misused. For example, the entry word “follow”
The rule indicates that if there is no preceding part of speech, or if it is a punctuation mark or a reading point, that is, if it is the beginning of a sentence or word, it is misused. Also, since it is better to write in hiragana regardless of the function in the sentence like “to”, “T” is put in the rule. Here, only the information of the morphemes before and after is used, but the information of other words may be used.

In the first example sentence, the case of "according to" is at the beginning of the sentence, so Pre. Part of speech = φ. As a result, the part of the rule becomes true, and "following" may be misused. On the other hand, since "aru" is the part of speech of the immediately preceding morpheme "de" is the connecting particle and the surface layer is "de", the rule part is false and it is judged not to be misused. If it is determined that there is a possibility of misuse, it can be displayed on the display unit 6 to inform the user of that fact, and an inquiry as to whether to rewrite can be made.

FIG. 6 is an explanatory diagram of an example of a rewriting inquiry screen. In FIG. 6, the original sentence, information about a word that may be misused, and information inquiring whether or not to rewrite are displayed. Here, with respect to the instruction given by the input unit 12 for rewriting,
The character type rewriting unit 11 rewrites the surface layer portion in the morpheme information storage unit 1. Regarding the rewriting method, the kana-kanji conversion section or the kanji-kana conversion section may be used without using the character type rewriting section 11. In this case, the rewrite word information of the misuse detection rule 5 is unnecessary. In the first example sentence, there is one word that may be misused, but when there are a plurality of words, as shown in FIG. 6, rewriting queries may be made one by one, or some may be put together. It can also be configured to query.

FIG. 7 is an explanatory diagram of the morpheme information after the rewriting of the first example sentence. When the rewriting instruction is given to “follow”, the character type rewriting unit 11 receives the rewriting word “following” in the misuse detection rule 5 from the misuse detection unit 4, and the Rewrite "follow" in the part to "follow". As a result, the morpheme information stored in the morpheme information storage unit 1 becomes as shown in FIGS.

Next, a case where the second example sentence "Hanako is delicious is there" is input will be described with reference to FIGS. Figure 8
Explanatory drawing of an example of the morpheme information of a 2nd example sentence, FIG.
5 is an explanatory diagram of an example of syntax information of an example sentence of FIG. The morpheme information as shown in FIG. 8 is stored in the morpheme information storage unit 1, and the syntax information as shown in FIG. 9 is stored in the syntax information storage unit 7. As in the case of the first sentence example, the sentence may be morphologically analyzed by the morphological analysis unit and input. Also,
The sentence may be syntactically analyzed and input by the syntactic analysis unit, or may be syntactically analyzed and input based on the morpheme information stored in the morpheme information storage unit 1. In FIG. 9, the syntax information is shown as a tree diagram, but various data formats can be used as the data format to be stored in the syntax information storage unit 7. The morpheme information and the syntactic information can be integrated.

First, as in the case of the first sentence example, the hiragana writing recommended word search unit 2 compares the hiragana writing recommended word in the second sentence example with the hiragana writing recommended word dictionary 3 for comparison. Find out. When the content shown in FIG. 4 is stored in the hiragana writing recommended word dictionary 3, “to” in the second sentence example is extracted as the hiragana writing recommended word.

Next, the misuse detection unit 4 checks whether or not the hiragana writing recommended word is misused by using the misuse detection rule 5. Here, three types of information, that is, morpheme information, syntactic information, and whether or not there is an indispensable case when using hiragana writing recommended words as synonyms, are used for checking. As for the morpheme information, as in the case of the first sentence example, information such as the part of speech and the surface layer of the preceding and following words is directly obtained from the morpheme information storage unit 1. The syntax information showing the function in the sentence is
Obtained from the syntax information storage unit 7. Information about the essential case is obtained by using the essential case search unit 8, the meaning type dictionary 9, the essential case dictionary 10, and the syntax information.

First, the indispensable case search unit 8 checks from the syntax information whether the searched recommended hiragana writing word can be a word. If there is no possibility of becoming a verb, the misuse detection unit 4 determines the possibility of misuse based only on the morpheme information and the syntax information. When the hiragana writing recommended word becomes a noun, the essential case when it becomes a noun is searched from the essential case dictionary.

FIG. 10 is an explanatory diagram of an example of the essential case dictionary 10. The essential case dictionary 10 is composed of headwords and dictionary contents, and the headwords and dictionary contents are related by a pointer. The dictionary contents are composed of the case required by the idiom, which is a headword, and the meaning type taken by the case. “HUM” and “ANI” in the figure indicate the meaning types of human and animal, respectively. For example, "say"
Indicates that the dictionary is "HUM + ga", and that it requires a Ga case, and means that a person is a Ga case. Similarly, "Sing" and "Swim" require the Ga case because the dictionary contents are "ANI + ga".
It means taking an animal as a moth. The dictionary content of "call" is "***", which means that the case is required and the meaning type that follows it can be anything.

By pattern matching the contents of the essential case dictionary 10 with the syntax information stored in the syntax information storage unit 7, it is checked whether or not there is an essential case. If there is an essential case and a semantic type to be taken as an essential case is specified, the semantic type dictionary 9 is used to check whether or not it is an appropriate essential case.

FIG. 11 is an explanatory diagram of an example of the meaning type dictionary 9. The meaning type dictionary 9 is composed of headwords and dictionary contents, which are associated with each other by pointers.
Dictionary contents "HUM", "ANI", "CON", "L"
“OC” and “ABS” indicate the meaning types of human, animal, concrete noun, place / time, and abstract noun, respectively. For example, "Hanako" and "Taro" are dictionary contents "HUM, AN
It is associated with “I, CON” and indicates that “Hanako” and “Taro” have semantic types of humans, animals, and concrete nouns. Similarly, “Pochi” indicates that the meaning type of animals and concrete nouns, and “Shinagawa” has the meaning type of place and time and the meaning type of abstract nouns.

When the essential case dictionary 10 is searched and the essential case is designated, the meaning class dictionary 9 is searched for by the word of the meaning class part. If the dictionary contents of the meaning type dictionary 9 satisfy the designation of the meaning type to be taken as an essential case, it is assumed that the essential case exists. Also, if there is no designation in what is taken as an indispensable case, it is assumed that the indispensable case exists when the indispensable case is found.

For example, when the essential case dictionary 10 and the meaning type dictionary 9 have the contents shown in FIG. 10 and FIG. 11, respectively, the second example sentence is searched for “toru” extracted as a recommended word in hiragana writing. Since "toru" has a possibility of becoming a verb, the essential case dictionary 10 is searched for by the original form "say" of the verb "it" of "toru". As a result of the search, it is found that the essential case is “HUM + is”. The moth case is present in the syntactic information. That is, it is a part of "Hanako" + "ga". Therefore, the meaning type of “Hanako” is checked using the meaning type dictionary 9. At this time, the meaning type of "Hanako" is "HUM, ANI, CO
Since it is “N” and has the meaning type “HUM”,
"Hanako" + "ga" becomes the essential case of "to", and the essential case of "to" exists.

FIG. 12 is an explanatory diagram of another example of the misuse detection rule. In this example, in addition to morpheme rules, syntax rules and essential case rules are provided as rules. The rules regarding morphemes are the same as the rules described in FIG. The syntax rules are
It shows the function of the word in the sentence. In addition, the rule regarding the essential case indicates whether or not the essential case exists. If all three of these rules are met, then there is a possibility of misuse. For example, the entry word “toru”
Indicates that the rule regarding the morpheme is “T”, so it is always better to rewrite. Further, since the rule regarding the syntax is “predicate”, it means that it is better to rewrite when “to” is “predicate”. Furthermore, since the rule regarding the indispensable case is “yes”, it is indicated that it is better to rewrite it when the indispensable case exists. The misuse detection unit checks using such misuse detection rules.

In the second sentence example, "toru" extracted as the recommended hiragana writing word has a morpheme rule "T", and it can be seen from the syntax information shown in FIG. 9 that it is a predicate. Further, the search of the essential case dictionary 10 and the meaning type dictionary 9 by the essential case search unit 8 reveals that the essential case exists as described above. Thus, the three rules are satisfied. This indicates that it is better to rewrite "to say" because "to" can be misused.

As a result, as in the case of the first sentence example, a rewriting inquiry is made to the user, and if a rewriting instruction is given, the surface layer portion of the morpheme information and the syntax information are rewritten. FIG. 13 is an explanatory diagram of the morpheme information after rewriting the second example sentence, and FIG. 14 is an explanatory diagram of the syntax information after rewriting the second example sentence. The column of the surface layer of “to” of the morpheme information shown in FIG. 8 is rewritten to “say”. In addition, “say” in the syntax information shown in FIG. 9 is rewritten as “say”. In this way, the second example sentence, "The restaurant where Hanako is delicious is right there." Is replaced with "I said", and the sentence "The restaurant where Hanako is delicious is right there." Can be obtained.

Similarly, when a sentence "The value of information has dramatically increased due to the advent of machines called computers." Is input, "said" in this sentence is "liked". Can be rewritten as This is because the misuse detection rule shown in FIG. 12 has the syntactic information of “said” as a connective word and does not have a G case retrieved from the essential case dictionary shown in FIG. This is because it is satisfied.

In the above-mentioned sentence example, the case where rewriting occurs is shown. However, if the condition of the misuse detection rule is not satisfied, rewriting is not presented even if the kanji notation is a recommended word in hiragana writing. In this way, even for recommended words in Hiragana writing, select an appropriate notation from Hiragana and Kanji notation based on the formal and semantic composition of sentences such as morphemes, syntaxes, essential cases, etc. It can be presented to the user when there is a possibility of misuse.

In the above-described embodiment, the hiragana writing recommended word has been described, but the same effect can be obtained even if the hiragana writing recommended word is used as the kanji writing recommended word. Further, the present invention can be applied not only to rewriting between hiragana and kanji, but also to rewriting between various character types such as hiragana and katakana, katakana and kanji, and the like. For example, in rewriting between katakana and kanji, "100 US" can be rewritten to "100 meters". Further, as a rewriting between the arithmetic number and the Chinese number, "1 example" can be rewritten to "one example". Further, rewriting rules between these various character types can be collectively configured. To apply to other character types, the information stored in the hiragana writing recommended word dictionary 3, the misuse detection rule 5, the meaning type dictionary 9, and the essential case dictionary 10 in the configuration shown in FIG. 1 may be changed.

[0041]

As is apparent from the above description, according to the present invention, it is possible to detect and rewrite a desired expression form for a word for which a certain expression form is recommended according to the function of the word in the sentence. Therefore, there is an effect that more appropriate support can be provided regarding writing, proofreading, and revision of sentences.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a text processing device of the present invention.

FIG. 2 is a flowchart for explaining an outline of an operation in one embodiment of the text processing device of the present invention.

FIG. 3 is an explanatory diagram of a morphological analysis result of a first example sentence.

FIG. 4 is an explanatory diagram of an example of a hiragana writing recommended word dictionary 3.

FIG. 5 is an explanatory diagram of an example of the misuse detection rule 5.

FIG. 6 is an explanatory diagram of an example of a rewriting inquiry screen.

FIG. 7 is an explanatory diagram of morpheme information after rewriting the first example sentence.

FIG. 8 is an explanatory diagram of an example of morpheme information of a second example sentence.

FIG. 9 is an explanatory diagram of an example of syntax information of a second example sentence.

10 is an explanatory diagram of an example of the essential case dictionary 10. FIG.

FIG. 11 is an explanatory diagram of an example of the meaning type dictionary 9.

FIG. 12 is an explanatory diagram of another example of the misuse detection rule.

FIG. 13 is an explanatory diagram of morpheme information after rewriting the second example sentence.

FIG. 14 is an explanatory diagram of syntax information after rewriting the second example sentence.

[Explanation of symbols]

1 Morphological information storage unit, 2 Hiragana writing recommended word search unit, 3 Hiragana writing recommended word dictionary, 4 Misuse detection unit, 5
Misuse detection rule, 6 output section, 7 syntax information storage section,
8 essential case search section, 9 meaning type dictionary, 10 essential case dictionary, 11 character type rewriting section, 12 input section.

Claims (3)

[Claims] 1. A first expression in a sentence processing device that detects a portion described in another expression format for a first expression recommended word in which information about a morpheme of a sentence is input and a first expression in a sentence is recommended. A first expression recommended word dictionary in which the expression of the first expression of the recommended word and the expression of another expression format are described, and the first expression recommended word from the information on the morphemes input using the first expression recommended word dictionary The first expression recommended word searching means for searching for the first expression recommended word and the knowledge of what expression format the first expression recommended word should be described based on the position of the first expression recommended word and the part of speech of words before and after. A sentence processing device comprising: an misuse detection rule holding unit that holds an misuse detection rule; and an misuse detection unit that detects misuse of a first expression recommended word using the information about the morpheme and the misuse detection rule. 2. The information on the syntax of the sentence is further input, and in the misuse detection rule, in addition to the position of the first expression recommended word and the part of speech of words before and after the first expression recommended word, the first expression recommended word may serve as a noun. Based on the presence / absence of the expression, knowledge of what expression format the first expression recommended word should be described is described,
The sentence processing apparatus according to claim 1, wherein the misuse detection unit detects misuse of the first recommended expression word by using information about a morpheme and syntax of an input sentence and the misuse detection rule. . 3. An essential case dictionary in which a case required when a first expression recommended word or the like is used as a noun is described, a meaning type dictionary describing a meaning type to which a word belongs, and the first expression When there is a possibility that the first expression recommended word searched by the recommended word searching means may act as a verb, the information about the syntax of the input sentence, the essential case dictionary, and the meaning type dictionary Is provided with an essential case search means for checking the existence of a word or phrase corresponding to the essential case, and in the misuse detection rule, the position of the first expression recommended word, the part of speech of words before and after, and the first recommended word may work as a noun. In addition to the presence / absence of the phrase, knowledge of what expression format the first expression recommended word should be described based on whether or not the first expression recommended word is indispensable when acting as a noun. The misuse detection means is a morpheme of the input sentence. 3. The sentence processing device according to claim 2, wherein the misuse of the first expression recommended word is detected by using the information about the information, the information about the syntax, the search result by the essential case search means, and the misuse detection rule.