JP2970443B2 - Document search device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document retrieval apparatus for retrieving a desired document from digitized and stored document information. The present invention relates to a retrieval of a database in which document information is stored, a general word processor, and an office. It can be used for searching for document information stored in a storage device such as a computer.

[0002]

2. Description of the Related Art In recent years, as a large amount of digitized document information such as e-mails, electronic catalogs, and electronic publications has started to be distributed in large quantities, a document search for searching only desired documents from the document information has been started. Interest in devices is growing. Conventionally, such a document search apparatus generally uses a technique called a keyword search using a keyword assigned to each document. However, a problem that the work of assigning a keyword cannot keep up with an increase in stored documents. The point is pointed out, and in recent years, a technique called full-text search, which solves the problem, has attracted attention. In the full-text search, keywords are assigned in advance in order to output a document that satisfies the search conditions by collating the search conditions based on the character string given by the user with all the character strings constituting the stored document. No need to do.

However, since the full-text search employs a simple search model that outputs all documents including a character string given as a search condition, the search condition is changed to a character string (or a logical expression of a plurality of character strings). However, there is a problem that it is difficult to accurately express and input various search intentions of the user.

Therefore, in order to alleviate such a problem in the full-text search, a language sentence has conventionally been allowed in the input,
A method of automatically extracting a search condition from the input sentence has been proposed (JP-A-1-185724).

[0005]

In the above-mentioned conventional method for automatically extracting a search condition from an input sentence, a predetermined word is extracted from the input sentence, and the extracted word is always extracted according to a thesaurus or a synonym / synonym according to a certain extended rule. The search conditions are expanded by using a dictionary or the like (the extracted words and their synonyms, synonyms, upper / lower concept words, etc. are ORed).

However, in this method, the extracted words are
Syntactic positioning of the word in the input sentence (for example,
Irrespective of the purpose case, place case, main part, predicate, etc.), that is, regardless of the context of the input sentence.

For this reason, in this method, there is a high possibility that the search condition will be extended far from the intention of the user, so that an appropriate search cannot be performed and the accuracy and efficiency of the search are not good.

[0008] The present invention has been made in view of the above circumstances, and it is possible to extend search conditions close to the user's intention based on an input sentence, thereby improving search accuracy and efficiency. It is an object to provide a document search device.

[0009]

In order to achieve this object, a document search apparatus according to a first aspect of the present invention analyzes an input sentence input by a user and analyzes a syntax pattern of the input sentence and the syntax pattern of the input sentence. A language processing unit that obtains an analysis result indicating an element of the input sentence according to a syntax pattern; an extension rule storage unit that stores, for each syntax pattern, an extension rule of an element of the input sentence according to the syntax pattern; A dictionary unit storing expansion contents according to each expansion rule for each element of the input sentence, based on the analysis result of the language processing unit, referring to the expansion rule storage unit and the dictionary unit,
An expansion unit that obtains expansion contents corresponding to the input sentence by expanding each element of the input sentence according to an expansion rule according to a syntax pattern of the input sentence, and an expansion unit that corresponds to the input sentence obtained by the expansion unit. The system includes a search condition generating unit that generates a search condition according to the expanded content, and a search unit that searches for a document according to the search condition.

A document search device according to a second aspect of the present invention analyzes an input sentence input by a user and analyzes the syntax pattern of the input sentence and the elements of the input sentence according to the syntax pattern. A language processing unit that obtains, an extension rule storage unit that stores, for each syntax pattern, an extension rule of an element of an input sentence according to the syntax pattern, and an extension rule for each element of the input sentence. A dictionary unit storing the extended contents according to the above, and based on the analysis result of the language processing unit, referring to the extension rule storage unit and the dictionary unit, each element of the input sentence is converted into a syntax pattern of the input sentence. An extension unit that obtains a candidate for an extension corresponding to the input sentence by extending according to the corresponding extension rule, and a candidate for an extension corresponding to the input sentence obtained by the extension unit are provided to the user. Means for performing, based on a candidate for extended content corresponding to the input sentence presented by the user, means for determining an extended content to be used for a search, and a search in accordance with the extended content determined by the user A search condition generating unit for generating a condition, and a search unit for searching for a document according to the search condition are provided.

A document search device according to a third aspect of the present invention receives a plurality of search requests from a user after a search start command is given by a user and before a search end command is given. In a document search apparatus in which an input sentence is input by a user at the time of inputting, the input sentence input at the time of each search request by the user is analyzed, and the syntax pattern of the input sentence and the input according to the syntax pattern A language processing unit that obtains an analysis result indicating a sentence element; an extension rule storage unit that stores, for each syntax pattern, an extension rule of an input sentence element according to the syntax pattern; On the other hand, based on the analysis result of the language processing unit, the dictionary unit storing the extension contents according to the respective extension rules, and referring to the extension rule storage unit and the dictionary unit, An extension unit that obtains a candidate for extended content corresponding to the input sentence by extending a prime according to an extension rule corresponding to a syntax pattern of the input sentence, and presents a user with a candidate for extended content corresponding to the input sentence to a user. Presenting means, means for deciding an extended content to be used for retrieval based on a candidate for extended content corresponding to the input sentence presented by the user, and retrieval according to the extended content decided by the user A search condition generation unit for generating a condition, a search unit for searching for a document in accordance with the search condition, and a search unit for receiving a search start command and a search end command. History analysis means for storing, as a history, an analysis result, a candidate for an extended content corresponding to the input sentence presented to the user, and an extended content determined by the user. A. And
For the first search request after the search start command, the presenting means presents to the user a candidate for an expansion content corresponding to the input sentence obtained by the expansion unit, and after the search start command, For the second and subsequent search requests, the extension corresponding to the input sentence obtained by referring to the history up to the previous time based on the candidate for the extension content corresponding to the input sentence obtained by the extension unit Content candidates are presented to the user.

A document search device according to a fourth aspect of the present invention is a document search device that receives a plurality of search requests from a plurality of users and inputs an input sentence from the user at each search request. A language processing unit that analyzes an input sentence input by a user to obtain a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern; and a syntax pattern for each syntax pattern. An extension rule storage unit that stores extension rules for elements of an input sentence according to the above, a dictionary unit that stores extension contents according to each extension rule for each element of the input sentence, and an analysis of the language processing unit Based on the result, by referring to the extension rule storage unit and the dictionary unit, each element of the input sentence is extended in accordance with an extension rule according to the syntax pattern of the input sentence, thereby expanding the input sentence. An extension unit for obtaining a candidate for an extended content corresponding to a sentence; a presentation unit for presenting a candidate for an extended content corresponding to the input sentence to a user; and an extended content corresponding to the input sentence presented by the user. Means for determining an extended content to be used for a search based on the candidate, a search condition generating unit for generating a search condition according to the extended content determined by the user, and a search for searching a document according to the search condition And a history of, for each user or each group to which the user belongs, the analysis result, a candidate for an extended content corresponding to the input sentence presented to the user, and an extended content determined by the user. And a history storage means for storing as Then, the presenting means presents to the user a candidate for an expansion content corresponding to the input sentence obtained by the expansion unit in response to a first search request by the user or a group to which the user belongs. And
For the second or subsequent search request by the user or the group to which the user belongs, based on the candidate for the expansion content corresponding to the input sentence obtained by the expansion unit, By referring to the history of the group to which the user belongs up to the previous time, a candidate for extended content corresponding to the obtained input sentence is presented to the user.

A document search device according to a fifth aspect of the present invention is a document search device in which a plurality of search requests are received from a plurality of users and an input sentence is input by the user at each search request. A language processing unit that analyzes an input sentence input by a user to obtain a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern; and a syntax pattern for each syntax pattern. An extension rule storage unit that stores extension rules for elements of an input sentence according to the above, a dictionary unit that stores extension contents according to each extension rule for each element of the input sentence, and an analysis of the language processing unit Based on the result, by referring to the extension rule storage unit and the dictionary unit, each element of the input sentence is extended in accordance with an extension rule according to the syntax pattern of the input sentence, thereby expanding the input sentence. An extension unit for obtaining a candidate for an extended content corresponding to a sentence; a presentation unit for presenting a candidate for an extended content corresponding to the input sentence to a user; and an extended content corresponding to the input sentence presented by the user. Means for determining an extended content to be used for a search based on the candidate, a search condition generating unit for generating a search condition according to the extended content determined by the user, and a search for searching a document according to the search condition And a history of, for each user or each group to which the user belongs, the analysis result, a candidate for an extended content corresponding to the input sentence presented to the user, and an extended content determined by the user. And a history storage means for storing as Then, the presenting means is based on a candidate for an expansion content corresponding to the input sentence obtained by the expansion unit, and the history of the user arbitrarily selected by the user or an arbitrary by the user. By referring to the history of the group of users selected in (1), candidates for extended contents corresponding to the obtained input sentence are presented to the users.

[0014]

According to the document retrieval apparatus of the first to fifth aspects of the present invention, the language processing unit obtains a syntax pattern of an input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern, and is extended. The section expands each element of the input sentence according to an expansion rule according to the syntax pattern of the input sentence, and obtains extended contents corresponding to the input sentence. Therefore, even if the element itself is the same, it depends on the syntactic position of the element in the input sentence (for example, object case, place case, main part, predicate, etc.), that is, according to the context of the input sentence. hand,
Can be extended according to appropriate extension rules. For this reason, search conditions that are close to the user's intention are expanded, so that a more appropriate search can be performed as compared with the related art, and search accuracy and efficiency can be improved.

According to the document retrieval apparatus according to the second to fifth aspects of the present invention, the extended content corresponding to the input sentence obtained by the extension unit is once presented to the user as a candidate, and the user is provided with the extended content. Can be determined, so that the user's search intention can be efficiently clarified and the document can be searched.

According to the third to fifth aspects of the present invention, the analysis result of the input sentence, the extended content candidate corresponding to the input sentence presented to the user, and the extended content candidate determined by the user. The extended contents are stored as a history, and when the extended contents corresponding to the input sentence are presented to the user, the history, that is, the extended method by the user can be reflected, so that the user's search intention is more efficiently performed. The document can be searched by clarifying it.

According to the document search apparatus of the third aspect of the present invention, the history can be stored in a range from the search start command to the search end command. In other words, the history is stored for each search session, and the history is stored for each session. Since the history is used, the user can use his or her extended technique in the session for multiple subsequent search requests in one session and reflect it in the presentation of extended content candidates. It can be carried out.

According to the document search device of the fourth aspect of the present invention, the history is stored for each user or each group to which the user belongs, and the history is used for each user or each group. The user or a user belonging to the group can perform an efficient search because the expansion method of the user or the user's group is reflected in the presentation of expansion content candidates.

According to the document search apparatus of the fifth aspect of the present invention, the history is stored for each user or each group to which the user belongs, and the history of the user or the group arbitrarily selected by a certain user is used. Since the user's selected user or group's expansion method is reflected in the presentation of candidates for expansion contents, the user who has a similar search method to his / her By selecting a group, an efficient search can be performed.

[0020]

【Example】

(Embodiment 1) Hereinafter, a document search apparatus according to a first embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a document search apparatus according to the first embodiment of the present invention.

This document search device, as shown in FIG.
An input / output device 11 including an output device such as a display device such as a CRT and an input device such as a keyboard and a mouse;
Language processing unit 13, extension unit 14, dictionary unit 15,
It is composed of an extension rule storage unit 16, a search condition generation unit 17, and a search unit 18. In FIG. 1, reference numeral 19 denotes a database in which documents to be searched are stored, and the database 19 includes the full text of the documents. Instead of the database 19, a storage unit storing a document in a word processor or the like may be used.

The user can use the input / output device 11 to input an input sentence, such as a natural language or a simple natural language, indicating the content of the search. The input / output device 11 is configured to display search results and the like.

The language processing unit 13 receives an input sentence input by a user using the input / output device 11 via the control unit 12 and uses a grammar and a dictionary (not shown) described in advance therein. The input sentence is analyzed to obtain an analysis result indicating a syntax pattern of the input sentence and elements of the input sentence according to the syntax pattern.

FIG. 2 shows an example of the syntax pattern. FIG. 2 shows only the syntax pattern of the general concept.
In this example, for example, as a syntax pattern of a subordinate concept of pattern 2 in FIG. 2, <qualifying part: relative clause <nominative: (empty)
Object case: X, predicate: Y {variant>, main part: noun (phrase): Z {material noun>, and the like. Here, the symbol “＠” indicates that the content on the left is specified as a lower concept by the content on the right. X, Y, Z
Are variables indicating the corresponding elements in the input sentence, but their description is omitted in pattern 4 and subsequent patterns in FIG.

Referring again to FIG. 1, the extension rule storage section 16 stores, for each syntax pattern, an extension rule of an element of an input sentence according to the syntax pattern.

FIG. 3 shows an example of an extension rule stored in the extension rule storage unit 16. For example, rule 2 in FIG. 3 indicates that when the modifier pattern of the syntax pattern is <relative clause <nominative case: (empty), objective case: X, predicate: Y >>,
The element X of the input sentence is extended as "different notation extension + synonymous extension + upper limit + lower limit", and the element Y of the input sentence
This is an extension rule indicating that an extension of "ignore" is to be performed. Here, the symbol “+” indicates that the logical sum of each expansion result is taken.

Referring to FIG. 1 again, the dictionary unit 15 stores expansion contents according to each expansion rule for each element of the input sentence. For example, the dictionary unit 15 includes a thesaurus, a synonym dictionary, a synonym dictionary, a related word dictionary, and the like.

The extension unit 14 receives the analysis result obtained by the language processing unit 13 via the control unit 12,
Based on the result of the analysis, referring to the extension rule storage unit 16 and the dictionary unit 15, each element of the input sentence is extended according to an extension rule corresponding to the syntax pattern of the input sentence, thereby corresponding to the input sentence. Get extended content.

The search condition generation unit 17 sends the expanded content corresponding to the input sentence obtained by the expansion unit 14 to the control unit 1.
2 and generates a search condition corresponding to the expansion content corresponding to the input sentence obtained by the expansion unit 14.
The search condition generation unit 17 outputs, for example,
Generates a logical expression of a character string. However, the search condition generation unit 17 refers to the analysis result and restricts not only the logical expression of the character string but also the order of the character string (for example, the character string A must be before the character string B). Various conditions may be generated as search conditions, such as restrictions on the intervals between character strings (for example, the interval between character strings A and B must be within 100 characters).

The search unit 18 receives the search condition generated by the search condition generation unit 17, searches the database 19 for a document satisfying the search condition, and supplies the search result to the control unit 12.

The control section 12 controls all the components described above. First, when receiving a search request and an input sentence indicating a search content after receiving a search start command from the input / output device 11 by a user operation, the control unit 12 supplies the input sentence to the language processing unit 13. Next, when receiving the analysis result from the language processing unit 13, the control unit 12 supplies the analysis result to the extension unit 14. Next, when receiving the extension content corresponding to the input sentence from the extension unit 14, the control unit 12 supplies the extension content to the search condition generation unit 17. Thereafter, when receiving the search result from the search unit 18, the control unit 12 controls the input / output device 11 to display the search result on the input / output device 11. If the user is not satisfied with the result of the search, the user operates the input / output device 11 to input a search request and a different input sentence, and repeats the search. However, the obtained search result is satisfactory. In such a case, a search end command is given from the input / output device 11. When receiving the search end command, the control unit 12
A predetermined search end process such as returning the display to the initial state is performed.

Here, as an example, specific operations of the language processing unit 13, the expansion unit 14, and the search condition generation unit 17 when an input sentence “device for searching a document” is input by a user will be described. I do.

In this case, the language processing unit 13 determines that the syntax pattern of the input sentence is a syntax pattern of a lower concept of the pattern 2 in FIG. 2 <modification part: relative clause <nominative: (empty),
X = “document”, Y = “search”, with the object case: X, the predicate: Y {variable>, the main part: noun (phrase): Z {material noun>, as elements of the input sentence according to this syntax pattern. ",
Z = “device” is obtained as an analysis result of the input sentence.

The extension unit 14 selects rules 2, 3 and 5 in FIG. 3, which are extension rules conforming to the syntax pattern, from the extension rule storage unit 16, and selects an element X of the input sentence = “document”. From the rule 2 and the rule 3, it is necessary to perform the extension “different notation extension + synonymous extension + upper limit + lower limit”. From the rule 3, it is necessary to perform the extension "different notation extension + synonymous extension" from rule 3. For the element Z of the input sentence = "device", the rule 5 is "different notation extension + synonymous extension + lower limit". "Should be extended. In this example, for element Y of the input sentence = “search”, “ignore” from rule 2
"Extended notation + Synonymous extension" from Rule 3
Is competing with the extension. In the case of such a conflict, which extension is to be performed is determined in advance. Here, the rule 3 is predetermined so as to give priority to the rule 3 (that is, in this example, the element Y of the input sentence is extended as “different notation extension + synonymous extension” for “search”). And

Then, the extension unit 14 actually expands each element of the input sentence with reference to the dictionary unit 15 according to such information, and obtains extended contents corresponding to the input sentence. That is,
The element “document” of the input sentence is extended as “document + document + text + book + book”, for example, by “different notation extension + synonymous extension + upper limit + lower limit”. The element “search” of the input sentence is extended as “different notation extension + synonymous extension”, for example, to “search + search + search + search”. The element “apparatus” of the input sentence is expanded as “different notation extension + synonymous extension + lower limit” and is extended to, for example, “apparatus + machine + processing apparatus”. here,
The symbol "+" indicates the logical sum of the preceding and following items.

The search condition generation unit 17 performs a search condition according to the extended content corresponding to such an input sentence, for example, "(document + document + text + book + book) * (search + search + search + search) * (Apparatus + machine + processing apparatus) ". Here, the symbol “*” indicates the logical product of the preceding and following items.

Next, as another example, when a user inputs an input sentence "murder in the metropolitan area", the language processing unit 13, the expansion unit 14, and the search condition generation unit 1
7 will be described in detail.

In this case, the language processing unit 13 determines that the syntax pattern of the input sentence is a syntax pattern of a lower concept of the pattern 1 in FIG. 2 <modification part: relative clause <nominative: (empty),
Place case: X, predicate: Y>, main part: noun (phrase): Z {abstract noun>, X = “metropolitan area”, Y = “occurred” as an element of an input sentence according to this syntax pattern, Z = “murder case” is obtained as an analysis result of this input sentence.

The extension unit 14 selects rules 1 and 6 in FIG. 3 which are extension rules conforming to the syntax pattern from the extension rule storage unit 16, and selects an element X of the input sentence = “metropolitan area”. From Rule 1, "Extended Notation + Synonymous Extension +
To be extended, "lower limit", element Y of input sentence
= For "occurred", the rule "Ignore" should be extended from rule 1. For input element Z = "murder case", rule 6 should be "different notation extension + synonymous extension + related word extension.""Should be extended.

Then, the extension unit 14 actually expands each element of the input sentence with reference to the dictionary unit 15 according to such information, and obtains extended contents corresponding to the input sentence. That is,
The input sentence element "Metropolitan area" is expressed as "different notation extension + synonymous extension +
An extension called "lower limit" is performed, and is extended to, for example, "Metropolitan area + Tokyo + Kanagawa + Chiba + Saitama + Gunma + Ibaraki + Yamanashi". The element "occurred" in the input sentence is expanded to "ignore" and this element is deleted. The element "murder case" of the input sentence is expanded to "expansion of different notation + synonym expansion + related word expansion", and is expanded to, for example, "murder case + killing case + evil case + evil crime".

The search condition generating unit 17 searches for a search condition according to the extension content corresponding to such an input sentence, for example, "(Metropolitan area + Tokyo + Kanagawa + Chiba + Saitama + Gunma + Ibaraki + Yamanashi) *
(Murder case + killing case + brutal case + brutal crime) ".

As described above, according to the present embodiment,
The language processing unit 13 obtains a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern, and the expansion unit 14 converts each element of the input sentence according to an expansion rule corresponding to the syntax pattern of the input sentence. Expanded to obtain the extended content corresponding to the input sentence. Therefore, even if the element itself is the same, it depends on the syntactic position of the element in the input sentence (for example, object case, place case, main part, predicate, etc.), that is, according to the context of the input sentence. And can be extended according to appropriate extension rules. For this reason,
As a result, the search conditions are expanded more closely to the user's intention, so that a more appropriate search can be performed as compared with the related art, and the accuracy and efficiency of the search can be improved.

(Embodiment 2) Hereinafter, a document search apparatus according to a second embodiment of the present invention will be described. Since the document search apparatus according to the present embodiment is basically the same as the document search apparatus according to the first embodiment, it will be described with reference to FIGS. 1 to 3 described above, and further, FIGS. This will be described with reference to FIG.

The difference between the present embodiment and the first embodiment will be described. Referring to FIG. 1, in the first embodiment, in the first embodiment, the control unit 12 controls the extension contents corresponding to the input sentence obtained by the extension unit 14. Is always supplied to the search condition generation unit 17 as it is, and the search condition generation unit 17 generates the search condition according to the expanded contents.
Is displayed on the input / output device 11 as a candidate for the extended content corresponding to the input sentence obtained by the extension unit 14 and presented to the user,
After the user operates the input / output device 11 to change, add, or delete the extended content as necessary, the extended content used for the search can be determined, and the search condition generating unit 17 is determined by the user. Generate search conditions according to the extended content.

Except for this point, the present embodiment is the same as the first embodiment, and the duplicate description is omitted.

Next, as an example, the operation of the document search apparatus according to the present embodiment when an input sentence “murder in the metropolitan area” is input by the user will be described.

First, upon receiving a search start command from the input / output device 11 by a user's operation, the control unit 12 receives a search request and an input sentence indicating a search content “murder in the metropolitan area”. The input sentence is supplied to the language processing unit 13.

The language processing section 13 analyzes the input sentence “murder in the metropolitan area” received from the control document search device 12 by using the grammar and the dictionary described in advance therein. The syntax pattern is a syntax pattern of a lower concept of pattern 1 in FIG. 2 <modification part: relative clause <nominative:
(Empty), place case: X, predicate: Y>, main part: noun (phrase):
Z = abstract noun> as elements of an input sentence according to this syntax pattern, X = “metropolitan area”, Y = “occurred”, Z =
The “murder case” is obtained as an analysis result of the input sentence.

Next, the control unit 12 receives the analysis result from the language processing unit 13 and supplies the analysis result to the extension unit 14.

The extension unit 14 selects the rules 1 and 6 in FIG. 3, which are extension rules conforming to the syntax pattern of the analysis result, from the extension rule storage unit 16 and sets the element X of the input sentence to “metropolitan area”. For Rule 1, "Different notation expansion +
An extension "synonymous extension + lower limit" should be performed, an input sentence element Y = "occurred" should be expanded from rule 1 to "ignore", an input sentence element Z = "murder case" , It is learned from rule 6 that an extension of “different notation extension + synonymous extension + related word extension” should be performed.

Then, the extension unit 14 actually extends each element of the input sentence with reference to the dictionary unit 15 according to such information, and obtains extended contents corresponding to the input sentence. That is,
The input sentence element "Metropolitan area" is expressed as "different notation extension + synonymous extension +
An extension called "lower limit" is performed, and is extended to, for example, "Metropolitan area + Tokyo + Kanagawa + Chiba + Saitama + Gunma + Ibaraki + Yamanashi". The element "occurred" in the input sentence is expanded to "ignore" and this element is deleted. The element "murder case" of the input sentence is expanded to "expansion of different notation + synonym expansion + related word expansion", and is expanded to, for example, "murder case + killing case + evil case + evil crime".

Next, the control unit 12 receives the extension contents corresponding to the input sentence obtained by the extension unit 14 and
1 as input and output device 1
1 is displayed as shown in FIG. 4 and presented to the user. FIG. 4 is a diagram illustrating a display example of candidates for extended contents corresponding to the input sentence “murder in the metropolitan area”.

The user looks at the presented extension content candidate and operates the input / output device 11 to change, add, or delete the extension content as necessary. The user operates the input / output device 11 according to his / her search intention, for example, operates the input / output device 11 to expand the “Metropolitan area” to “Metropolitan area”,
Select only "Tokyo", "Kanagawa", "Chiba" and "Saitama", and select only "murder", "murder" and "evil crime" for "murder". The control unit 12
In response to such a selection, the input / output device 11 is controlled to display, for example, a display as shown in FIG. The underline in FIG. 5 indicates the extension selected by the user.
Then, when the user selects the extension content that the user intends, the user operates the input / output device 11 to input the fact,
The extension content is determined. In the example described above, an example of selecting an extension content candidate is used. However, the user may confirm the extension content candidate as it is, or add the extension content (for example, in the “Metropolitan area”). After adding “Shizuoka”, the extended content may be determined.

Next, the control unit 12 supplies the expanded condition corresponding to the input sentence thus determined by the user to the search condition generation unit 17.

The search condition generation unit 17 receives from the control unit 12 the extended contents corresponding to the input sentence determined by the user, and retrieves the search conditions according to the extended contents, for example, “(Metropolitan area + Tokyo + Kanagawa + (Chiba + Saitama) * (murder case + killing case + violent crime) ".

The search unit 18 receives the search condition generated by the search condition generation unit 17, searches the database 19 for a document satisfying the search condition, and supplies the search result to the control unit 12.

Upon receiving the search result from the search unit 18, the control unit 12 controls the input / output device 11 and
1 and the search result is displayed. If the user is not satisfied with the result of the search, the user operates the input / output device 11 to input a search request and a different input sentence, and repeats the search. However, the obtained search result is satisfactory. In such a case, a search end command is given from the input / output device 11. When receiving the search end command, the control unit 12 performs a predetermined search end process such as returning the display of the input / output device 11 to an initial state.

As described above, according to the present embodiment,
As in the first embodiment, the language processing unit 13 obtains the syntax pattern of the input sentence and the analysis result indicating the elements of the input sentence according to the syntax pattern. The extension is performed according to the extension rule according to the syntax pattern of the input sentence, and the extension content corresponding to the input sentence is obtained. Therefore, even if the element itself is the same, it depends on the syntactic position of the element in the input sentence (for example, object case, place case, main part, predicate, etc.), that is, according to the context of the input sentence. And can be extended according to appropriate extension rules. For this reason, search conditions that are close to the user's intention are expanded, so that a more appropriate search can be performed as compared with the related art, and search accuracy and efficiency can be improved.

Further, according to the document search apparatus of the present embodiment, the extended content corresponding to the input sentence obtained by the extension unit 14 is once presented to the user as a candidate, and the user determines the extended content. Therefore, it is possible to efficiently clarify the user's search intention and perform a document search.

(Embodiment 3) Hereinafter, a document search apparatus according to a third embodiment of the present invention will be described with reference to the drawings.

FIG. 6 is a block diagram showing the configuration of a document search device according to the third embodiment of the present invention. 6, the same reference numerals are given to the same or corresponding components as those shown in FIG. 1 described above.

The document search device according to the present embodiment is provided with the second
It differs from the document search apparatus according to the embodiment in the following points.

That is, in this embodiment, a history storage unit 41 is added as shown in FIG. Further, in this embodiment, the control unit 12 clears the contents of the history storage unit 41 each time a search start command is given from the user by the input / output device 11, and the search start command is sent from the user by the input / output device 11. The analysis result obtained by the language processing unit 13 in response to each search request from when the search request is given until the search end command is given, a candidate for an extended content corresponding to the input sentence presented to the user, and the user The expansion content corresponding to the input sentence determined by the above is stored in the history storage unit 41 as a history. Furthermore, in the second embodiment, the control unit 12 always displays the extension contents corresponding to the input sentence obtained by the extension unit 14 as candidates as they are on the input / output device 11 and presents them to the user. On the other hand, in the present embodiment, the control unit 12 displays, on the input / output device 11, an extension content candidate corresponding to the input sentence obtained by the extension unit 14 in response to the first search request after the search start command. In response to the second or subsequent search request after the search start command, the history storage unit based on the candidate for the expansion content corresponding to the input sentence obtained by the expansion unit 14 and the previous storage unit The candidate of the extended content corresponding to the input sentence is obtained by referring to the history stored in 41, and the candidate of the extended content corresponding to the input sentence thus obtained is displayed on the input / output device 11 so that the user Present.

Except for this point, the present embodiment is the same as the second embodiment, and the duplicate description will be omitted.

Next, as an example, at the time of the first search request from the user after the search start command, an input sentence “murder in the Tokyo metropolitan area” is input, and two words after the search start command are input.
An operation of the document search device according to the present embodiment when an input sentence “traffic accident in the metropolitan area” is input at the time of the second search request will be described.

First, after receiving a search start command from the input / output device 11 by a user's operation, the control unit 12 receives the first search request and the input sentence “murder in the metropolitan area” indicating the search content. The input sentence is supplied to the language processing unit 13. The control unit 12 clears the contents of the history storage unit 41 when receiving the search request.

The language processing unit 13 analyzes the input sentence “murder in the metropolitan area” received from the control document search device 12 by using the grammar and the dictionary described in advance therein. The syntax pattern is a syntax pattern of a lower concept of pattern 1 in FIG. 2 <modification part: relative clause <nominative:
(Empty), place case: X, predicate: Y>, main part: noun (phrase):
Z = abstract noun> as elements of an input sentence according to this syntax pattern, X = “metropolitan area”, Y = “occurred”, Z =
The “murder case” is obtained as an analysis result of the input sentence.

Next, the control unit 12 receives the analysis result from the language processing unit 13 and supplies the analysis result to the extension unit 14.

The extension unit 14 selects the rules 1 and 6 in FIG. 3 which are extension rules conforming to the syntax pattern of the analysis result from the extension rule storage unit 16 and sets the element X of the input sentence to “metropolitan area”. For Rule 1, "Different notation expansion +
An extension "synonymous extension + lower limit" should be performed, an input sentence element Y = "occurred" should be expanded from rule 1 to "ignore", an input sentence element Z = "murder case" , It is learned from rule 6 that an extension of “different notation extension + synonymous extension + related word extension” should be performed.

The extension unit 14 actually extends each element of the input sentence with reference to the dictionary unit 15 in accordance with such information, and obtains extended contents corresponding to the input sentence. That is,
The input sentence element "Metropolitan area" is expressed as "different notation extension + synonymous extension +
An extension called "lower limit" is performed, and is extended to, for example, "Metropolitan area + Tokyo + Kanagawa + Chiba + Saitama + Gunma + Ibaraki + Yamanashi". The element "occurred" in the input sentence is expanded to "ignore" and this element is deleted. The element "murder case" of the input sentence is expanded to "expansion of different notation + synonym expansion + related word expansion", and is expanded to, for example, "murder case + killing case + evil case + evil crime".

Next, the control unit 12 receives the extension contents corresponding to the input sentence obtained by the extension unit 14 and
1 as input and output device 1
1 is displayed as shown in FIG. 4 and presented to the user. FIG. 4 is a diagram illustrating a display example of candidates for extended contents corresponding to the input sentence “murder in the metropolitan area”.

The user looks at the presented extension content candidate and operates the input / output device 11 to change, add or delete the extension content as necessary. The user operates the input / output device 11 according to his / her search intention, for example, operates the input / output device 11 to expand the “Metropolitan area” to “Metropolitan area”,
Select only "Tokyo", "Kanagawa", "Chiba" and "Saitama", and select only "murder", "murder" and "evil crime" for "murder". The control unit 12
In response to such a selection, the input / output device 11 is controlled to display, for example, a display as shown in FIG. The underline in FIG. 5 indicates the extension selected by the user.
Then, when the user selects the extension content that the user intends, the user operates the input / output device 11 to input the fact,
The extension content is determined.

Next, the control unit 12 supplies the extended content corresponding to the input sentence thus determined by the user to the search condition generation unit 17. In response to the current search request, the control unit 12 determines the analysis result obtained by the language processing unit 13, a candidate for an extended content corresponding to the input sentence presented to the user, and the user. The extended content corresponding to the input sentence is stored as a history.

Next, the search condition generation unit 17 receives the extended content corresponding to the input sentence determined by the user from the control unit 12, and retrieves the search condition according to the extended content, for example, “(Metropolitan area + Tokyo + Kanagawa + Chiba + Saitama) * (murder case + killing case + evil crime).

The search unit 18 receives the search condition generated by the search condition generation unit 17, searches the database 19 for a document satisfying the search condition, and supplies the search result to the control unit 12.

When receiving the search result from the search unit 18, the control unit 12 controls the input / output device 11 and
1 shows the search result.

Next, without receiving a search end command from the input / output device 11 by the user, the user receives a second search request after the search start command and an input sentence "traffic accident in the metropolitan area" indicating the search content. And the input sentence
To supply.

The language processing unit 13 analyzes the input sentence “traffic accident in the metropolitan area” received from the control document search device 12 using the grammar and the dictionary described in advance therein, and analyzes the input sentence. The syntax pattern is a syntax pattern of a lower concept of pattern 1 in FIG. 2 <modification part: relative clause <nominative:
(Empty), place case: X, predicate: Y>, main part: noun (phrase):
As an element of an input sentence according to this syntax pattern, X = “metropolitan area”, Y = “occurred”, and Z = “traffic accident” are obtained as analysis results of the input sentence.

Next, the control unit 12 receives the analysis result from the language processing unit 13 and supplies the analysis result to the extension unit 14.

The extension unit 14 selects the rules 1 and 6 in FIG. 3 which are extension rules conforming to the syntax pattern of the analysis result from the extension rule storage unit 16 and sets the element X of the input sentence to “metropolitan area”. For Rule 1, "Different notation expansion +
An extension of "synonymous extension + lower limit" should be performed, an extension of "ignore" should be performed from rule 1 for element Y = "occurred" of input sentence, an element Z of input sentence should be "traffic accident" , It is learned from rule 6 that an extension of “different notation extension + synonymous extension + related word extension” should be performed.

Then, the extension unit 14 actually extends each element of the input sentence with reference to the dictionary unit 15 according to such information, and obtains extended contents corresponding to the input sentence. That is,
The input sentence element "Metropolitan area" is expressed as "different notation extension + synonymous extension +
An extension called "lower limit" is performed, and is extended to, for example, "Metropolitan area + Tokyo + Kanagawa + Chiba + Saitama + Gunma + Ibaraki + Yamanashi". The element “occurred” in the input sentence is expanded to “ignore” and this element is deleted. The element "traffic accident" of the input sentence is expanded as "different notation expansion + synonym expansion + related word expansion", for example, "traffic accident + rear-end collision + collision"
Is extended to

Next, since the search request is the second search request, the control unit 12 sets the history based on the candidate of the expansion content corresponding to the input sentence obtained by the expansion unit 14 and the history up to the previous time. With reference to the history stored in the storage unit 41, a candidate for an expansion content corresponding to the input sentence, which reflects the method of expanding the input sentence by the user up to the previous time, is obtained. In this example, as a related history, the history storage unit 41 stores “Metropolitan area + Tokyo +
While Kanagawa + Chiba + Saitama + Gunma + Ibaraki + Yamanashi was presented to the user, the user was informed that “Metropolitan area + Tokyo + Kanagawa + Chiba + Saitama” was confirmed as an expanded content to be used in the search. Since the history is stored as a history related to the current processing, the control unit 12 expands the element “metropolitan area” of the input sentence into “metropolitan area + Tokyo + Kanagawa + Chiba + Saitama” and the element “ The expanded content is obtained by deleting "occurred" and expanding the element "traffic accident" of the input sentence into "traffic accident + rear-end collision".

Then, the control unit 12 controls the input / output device 11 to store the extended content corresponding to the input sentence thus obtained,
The candidate is displayed on the input / output device 11 as shown in FIG. 7 and presented to the user. FIG. 7 is a diagram showing a display example of expansion content candidates corresponding to the input sentence "traffic accident in the metropolitan area" in this example.

The user looks at the presented extension content candidate and operates the input / output device 11 to change, add, or delete the extension content as necessary. The user operates the input / output device 11, for example, according to his or her search intention, and selects a candidate for the extended content corresponding to the presented input sentence as it is as the extended content corresponding to the input sentence used for the search. The control unit 12 controls the input / output device 11 in accordance with such a selection to cause a display as shown in FIG. 8 to be performed, for example. The underline in FIG. 8 indicates the extended content selected by the user. Then, when the user selects the extension content intended by the user, the user operates the input / output device 11 to input that fact, and the extension content is determined.

Next, the control unit 12 supplies the expanded content corresponding to the input sentence thus determined by the user to the search condition generation unit 17. In response to the current search request, the control unit 12 determines the analysis result obtained by the language processing unit 13, a candidate for an extended content corresponding to the input sentence presented to the user, and the user. The extended content corresponding to the input sentence is stored as a history.

Next, the search condition generation unit 17 receives the extended content corresponding to the input sentence determined by the user from the control unit 12, and retrieves the search condition according to the extended content, for example, “(Metropolitan area + Tokyo + Kanagawa + Chiba + Saitama) * (traffic accident + rear-end collision + collision).

The search unit 18 receives the search condition generated by the search condition generation unit 17, searches the database 19 for a document satisfying the search condition, and supplies the search result to the control unit 12.

When receiving the search result from the search unit 18, the control unit 12 controls the input / output device 11 and
1 shows the search result. If the user is not satisfied with the result of the search, the user operates the input / output device 11 to input a search request and a different input sentence, and repeats the search. However, the obtained search result is satisfactory. In such a case, a search end command is given from the input / output device 11. When receiving the search end command, the control unit 12 performs a predetermined search end process such as returning the display of the input / output device 11 to an initial state.

As described above, according to the present embodiment,
As in the first and second embodiments, the language processing unit 1
3 obtains a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern. The expansion unit 14 expands each element of the input sentence according to an expansion rule corresponding to the syntax pattern of the input sentence. The extended content corresponding to the input sentence is obtained. Therefore, even if the element itself is the same, it depends on the syntactic position of the element in the input sentence (for example, object case, place case, main part, predicate, etc.), that is, according to the context of the input sentence. And can be extended according to appropriate extension rules. For this reason, search conditions that are close to the user's intention are expanded, so that a more appropriate search can be performed as compared with the related art, and search accuracy and efficiency can be improved.

Further, according to the document search apparatus of the present embodiment, the extended content corresponding to the input sentence obtained by the extension unit 14 is once presented to the user as a candidate, and the user determines the extended content. Therefore, it is possible to efficiently clarify the user's search intention and perform a document search.

Further, according to the document search apparatus of the present embodiment, the history can be stored in a range from the search start command to the search end command. In other words, the history is stored for each search session, and the history is stored for each session. Since the history is used, the user performs an efficient search because the user's own extension method in the session is reflected in the presentation of the candidate of the extension content for a plurality of subsequent search requests in one session. be able to.

(Embodiment 4) Hereinafter, a fourth embodiment of the present invention will be described with reference to the drawings. The document search apparatus according to the present embodiment is basically the same as the document search apparatus according to the third embodiment, and will be described with reference to FIG.

A description will be given of points of the present embodiment which are different from the third embodiment. In FIG. 6, in the third embodiment, the control unit 12 clears the contents of the history storage unit 41 every time a search start command is given by the user from the input / output device 11, and In response to each search request from when the search start command is given until the search end command is given, the analysis result obtained by the language processing unit 13 and the expanded contents corresponding to the input sentence presented to the user The candidate and the extended content corresponding to the input sentence confirmed by the user are stored in the history storage unit 41 as a history. Further, in the third embodiment, in response to the first search request after the search start command, the control unit 12 outputs a candidate for expansion content corresponding to the input sentence obtained by the expansion unit 14 to the input / output device. 11 and presented to the user, and for the second and subsequent search requests after the search start command, based on the candidate for the expansion content corresponding to the input sentence obtained by the expansion unit 14, By referring to the history stored in the history storage unit 41, a candidate for an extended content corresponding to the input sentence is obtained, and a candidate for an extended content corresponding to the input sentence thus obtained is displayed on the input / output device 11. Was presented to the user.

On the other hand, in the present embodiment, the control device 1
2 does not clear the contents of the history storage unit 41 even when a search start command is given from the user through the input / output device 11, and the user or the group to which the user belongs when the search start command is given by the user. Receives a user ID or a group ID for identifying the user. In addition, for each of the user or each group to which the user belongs, the analysis result obtained by the language processing unit 13, the candidate of the extended content corresponding to the input sentence presented to the user, and the input determined by the user. The expansion content corresponding to the sentence is stored in the history storage unit 41 as a history. Further, in the present embodiment, the control unit 12 inputs a candidate for the expansion content corresponding to the input sentence obtained by the expansion unit 14 in response to the first search request by the user or the group to which the user belongs. It is displayed on the output device 11 and presented to the user. For the second or subsequent search request by the user or the group to which the user belongs, the expansion content corresponding to the input sentence obtained by the expansion unit 14 is displayed. Based on the candidate, and by referring to the history stored in the history storage unit 41 up to the previous time by the user or the group to which the user belongs, the candidate of the extended content corresponding to the input sentence is obtained. The candidate of the expansion content corresponding to the input sentence obtained as described above is displayed on the input / output device 11 and presented to the user.

That is, in the third embodiment, the history is stored for each session of the search, and the history is used for each session. In this embodiment, however, the history is not related to the session of the search. The history is stored for each user or each group to which the user belongs, and the history is used for each user or each group to which the user belongs.

Except for this point, the present embodiment is the same as the third embodiment, and the duplicate description will be omitted.

According to the present embodiment, the above-described first to third embodiments
In the same way as in the embodiment, the language processing unit 13 obtains a syntax pattern of the input sentence and an analysis result indicating the elements of the input sentence according to the syntax pattern, and the extension unit 14 converts the elements of the input sentence into the syntax of the input sentence. Extended contents corresponding to the input sentence are obtained by being expanded according to the expansion rule according to the pattern. Therefore, even if the element itself is the same, it depends on the syntactic position of the element in the input sentence (for example, object case, place case, main part, predicate, etc.), that is, according to the context of the input sentence. And can be extended according to appropriate extension rules. For this reason, search conditions that are close to the user's intention are expanded, so that a more appropriate search can be performed as compared with the related art, and search accuracy and efficiency can be improved.

Further, according to the document search apparatus of the present embodiment, the extended content corresponding to the input sentence obtained by the extension unit 14 is once presented to the user as a candidate, and the user determines the extended content. Therefore, it is possible to efficiently clarify the user's search intention and perform a document search.

Further, according to the document search apparatus of the present embodiment, the history is stored for each user or each group to which the user belongs, and the history is used for each user or each group. Alternatively, the user belonging to the group can perform an efficient search because the extension method of the user or his group is reflected in the presentation of the candidate of the expansion content.

Embodiment 5 Hereinafter, a fifth embodiment of the present invention will be described with reference to the drawings. The document search apparatus according to the present embodiment is basically the same as the document search apparatus according to the third embodiment, and will be described with reference to FIG.

A description will be given of the points of this embodiment different from the third embodiment. In FIG. 6, in the third embodiment, the control unit 12 clears the contents of the history storage unit 41 every time a search start command is given by the user from the input / output device 11, and In response to each search request from when the search start command is given until the search end command is given, the analysis result obtained by the language processing unit 13 and the expanded contents corresponding to the input sentence presented to the user The candidate and the extended content corresponding to the input sentence confirmed by the user are stored in the history storage unit 41 as a history. Further, in the third embodiment, in response to the first search request after the search start command, the control unit 12 outputs a candidate for expansion content corresponding to the input sentence obtained by the expansion unit 14 to the input / output device. 11 and presented to the user, and for the second and subsequent search requests after the search start command, based on the candidate for the expansion content corresponding to the input sentence obtained by the expansion unit 14, By referring to the history stored in the history storage unit 41, a candidate for an extended content corresponding to the input sentence is obtained, and a candidate for an extended content corresponding to the input sentence thus obtained is displayed on the input / output device 11. Was presented to the user.

On the other hand, in this embodiment, the control device 1
2 does not clear the contents of the history storage unit 41 even when a search start command is given from the user through the input / output device 11, and the user or the group to which the user belongs when the search start command is given by the user. Receives a user ID or a group ID for identifying the user. In addition, for each of the user or each group to which the user belongs, the analysis result obtained by the language processing unit 13, the candidate of the extended content corresponding to the input sentence presented to the user, and the input determined by the user. The expansion content corresponding to the sentence is stored in the history storage unit 41 as a history. Further, in the present embodiment, the control unit 12 stores the user arbitrarily selected by the user stored in the history storage unit 41 (when the input / output device 11 inputs a search request in advance, the user Refer to the history of a user group arbitrarily selected by the user (specify a group ID by the input / output device 11 in advance when a search request is input). Thus, a candidate for the extension content corresponding to the input sentence is obtained, and the candidate for the extension content is displayed on the input / output device 11 and presented to the user.

That is, in the third embodiment, the history is stored for each session of the search, and the history is used for each session. , The history is stored for each user or group to which the user belongs, and the history of the user or group arbitrarily selected by a certain user is used.

Except for this point, the present embodiment is the same as the third embodiment, and the duplicate description will be omitted.

According to the present embodiment, the above-described first to third embodiments
In the same manner as in the first embodiment, the language processing unit 13 obtains a syntax pattern of an input sentence and an analysis result indicating elements of the input sentence according to the syntax pattern. Extended contents corresponding to the input sentence are obtained by being expanded according to the expansion rule according to the pattern. Therefore, even if the element itself is the same, it depends on the syntactic position of the element in the input sentence (for example, object case, place case, main part, predicate, etc.), that is, according to the context of the input sentence. And can be extended according to appropriate extension rules. For this reason, search conditions that are close to the user's intention are expanded, so that a more appropriate search can be performed as compared with the related art, and search accuracy and efficiency can be improved.

Further, according to the document search apparatus of this embodiment, the extended content corresponding to the input sentence obtained by the extension unit 14 is once presented to the user as a candidate, and the user determines the extended content. Therefore, it is possible to efficiently clarify the user's search intention and perform a document search.

Further, according to the document search apparatus of this embodiment, the history is stored for each user or each group to which the user belongs, and the history of the user or the group arbitrarily selected by a certain user is used. Therefore, the user
Since the extension method of the selected user or group is reflected in the presentation of the candidate for the extension content, the user or group whose search method is similar to that of the user or group when there is no user ID and there is no history. By selecting, an efficient search can be performed.

Although the first to fourth embodiments of the present embodiment have been described above, the present invention is not limited to these embodiments.

For example, according to the third embodiment, in the third embodiment, in response to a subsequent search request of a certain user, the control unit 12 always sets expansion contents corresponding to the input sentence obtained by the expansion unit 14 as candidates. It is displayed on the input / output device 11 as it is and presented to the user, and at the same time, the content of the corresponding history is simply presented to the user. It may be only for the convenience of the user.

Further, in the fifth embodiment, when referring to the history of a certain user group when there is no user ID and no own history, one of the groups is selected and Although it is possible to refer only to the history of a person, it is also possible to perform processing that reflects the average preference of a person by referring to the average of the history (extended method) of the entire group.

[0112]

As described above, according to the present invention,
The language processing unit obtains a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern, and the expansion unit expands each element of the input sentence according to an extension rule according to the syntax pattern of the input sentence. Thus, the extended contents corresponding to the input sentence can be obtained. Therefore, even if the element itself is the same, it depends on the syntactic position of the element in the input sentence (for example, object case, place case, main part, predicate, etc.), that is, according to the context of the input sentence. And can be extended according to appropriate extension rules. For this reason, according to the present invention, search conditions that are close to the user's intention are expanded, so that a more appropriate search can be performed as compared with the related art, and search accuracy and efficiency can be improved. it can.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a document search device according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of the contents of a syntax pattern

FIG. 3 is a diagram showing an example of the contents of an extension rule.

FIG. 4 is a diagram showing a display example of extension content candidates corresponding to an input sentence;

FIG. 5 is a diagram showing a display example of a selection status of a candidate for an extended content shown in FIG. 4;

FIG. 6 is a block diagram showing a configuration of a document search device according to another embodiment of the present invention.

FIG. 7 is a diagram showing a display example of another extension content candidate corresponding to an input sentence;

FIG. 8 is a diagram showing a display example of the selection status of the extension content candidates shown in FIG. 7;

[Explanation of symbols]

 Reference Signs List 11 input / output device 12 control unit 13 language processing unit 14 extension unit 15 dictionary unit 16 extension rule storage unit 17 search condition generation unit 18 search unit 19 database 41 history storage unit

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-2-253474 (JP, A) JP-A-3-102566 (JP, A) JP-A-3-229366 (JP, A) JP-A-63- 219034 (JP, A) JP-A-4-237369 (JP, A) JP-A-1-185724 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 17/30

Claims (5)

(57) [Claims] A language processing unit for analyzing an input sentence input by a user to obtain a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern; An extension rule storage unit storing extension rules of elements of the input sentence according to the syntax pattern, a dictionary unit storing extension contents according to the extension rules for each element of the input sentence, The input is performed by expanding each element of the input sentence according to an expansion rule according to a syntax pattern of the input sentence with reference to the extension rule storage unit and the dictionary unit based on the analysis result of the language processing unit. An extension unit for obtaining an extension corresponding to the sentence; a search condition generation unit for generating a search condition corresponding to the extension corresponding to the input sentence obtained by the extension unit; Document search apparatus characterized by comprising a, a search unit for searching for documents Te. 2. A language processing unit for analyzing an input sentence input by a user to obtain a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern; An extension rule storage unit storing extension rules of elements of the input sentence according to the syntax pattern, a dictionary unit storing extension contents according to the extension rules for each element of the input sentence, The input is performed by expanding each element of the input sentence according to an expansion rule according to a syntax pattern of the input sentence with reference to the extension rule storage unit and the dictionary unit based on the analysis result of the language processing unit. An extension unit for obtaining a candidate for extension content corresponding to a sentence; a unit for presenting a candidate for an extension content corresponding to the input sentence obtained by the extension unit to a user; and the user is presented. Means for determining an extended content to be used for a search based on a candidate for an extended content corresponding to the input sentence; a search condition generating unit for generating a search condition according to the extended content determined by the user; A document search device comprising: a search unit that searches for a document according to a condition. 3. A search request is received from the user a plurality of times from when a search start command is given by the user to when a search end command is given, and an input sentence is input by the user at each search request. Processing for analyzing an input sentence input by a user at the time of each search request to obtain a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern Part, an extension rule storage unit storing, for each syntax pattern, an extension rule of an element of an input sentence according to the syntax pattern, and an extension content according to each extension rule for each element of the input sentence A dictionary unit storing the input sentence, and referring to the extension rule storage unit and the dictionary unit based on an analysis result of the language processing unit, according to a syntax pattern of the input sentence. An extension unit that obtains a candidate for an extension content corresponding to the input sentence by extending according to an extension rule; a presentation unit that presents a candidate for an extension content corresponding to the input sentence to a user; and the user is presented. Means for deciding an extended content to be used for a search based on extended content candidates corresponding to the input sentence, a search condition generating unit for generating a search condition according to the extended content decided by the user, A search unit that searches for a document according to search conditions; and, in response to each search request from when the search start command is given to when the search end command is given, the analysis result is presented to the user. A history storage unit that stores, as a history, a candidate for an extension content corresponding to the input sentence, and the extension content determined by the user, the presentation unit after the search start command For the first search request, a candidate for the expansion content corresponding to the input sentence obtained by the expansion unit is presented to the user, and for the second and subsequent search requests after the search start command, Presenting to the user a candidate for an expansion content corresponding to the input sentence obtained by referring to the history up to the previous time based on a candidate for an expansion content corresponding to the input sentence obtained by the expansion unit, A document search device characterized by the following: 4. A document search device, which receives a plurality of search requests from a plurality of users and receives an input sentence from the user at the time of each search request, analyzes the input sentence input from the user. A language processing unit for obtaining a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern; and for each syntax pattern, an extension rule of an element of the input sentence according to the syntax pattern. A stored expansion rule storage unit, a dictionary unit storing expansion contents according to each expansion rule for each element of the input sentence, and based on an analysis result of the language processing unit, the expansion rule storage unit and With reference to the dictionary unit, each element of the input sentence is expanded in accordance with an expansion rule according to the syntax pattern of the input sentence, thereby obtaining an extension content candidate corresponding to the input sentence. Unit, a presenting means for presenting a candidate for an extended content corresponding to the input sentence to a user, and an extended content used by the user for a search based on the candidate for an extended content corresponding to the presented input sentence A search condition generating unit that generates a search condition according to the extended content determined by the user; a search unit that searches for a document according to the search condition; and the user or the user For each group to which it belongs, the analysis result, a candidate for an extension content corresponding to the input sentence presented to the user, and history storage means for storing the extension content determined by the user as a history, The presenting means, for a first search request by the user or a group to which the user belongs, generates a candidate for expansion content corresponding to the input sentence obtained by the expansion unit. In response to the second or subsequent search request by the user or the group to which the user belongs, based on the candidate of the expansion content corresponding to the input sentence obtained by the expansion unit, and A document search device that presents to the user a candidate for extended content corresponding to the obtained input sentence by referring to the history of the user or the group to which the user belongs up to the previous time. 5. A document search apparatus in which a plurality of search requests are received from a plurality of users and an input sentence is input from the user at each search request, wherein the input sentence input by the user is analyzed. A language processing unit for obtaining a syntax pattern of the input sentence and an analysis result indicating an element of the input sentence according to the syntax pattern; and for each syntax pattern, an extension rule of an element of the input sentence according to the syntax pattern. A stored expansion rule storage unit, a dictionary unit storing expansion contents according to each expansion rule for each element of the input sentence, and based on an analysis result of the language processing unit, the expansion rule storage unit and With reference to the dictionary unit, each element of the input sentence is expanded in accordance with an expansion rule according to the syntax pattern of the input sentence, thereby obtaining an extension content candidate corresponding to the input sentence. Unit, a presenting means for presenting a candidate for an extended content corresponding to the input sentence to a user, and an extended content used by the user for a search based on the candidate for an extended content corresponding to the presented input sentence A search condition generating unit that generates a search condition according to the extended content determined by the user; a search unit that searches for a document according to the search condition; and the user or the user For each group to which it belongs, the analysis result, a candidate for an extension content corresponding to the input sentence presented to the user, and history storage means for storing the extension content determined by the user as a history, The presenting means is based on a candidate for expansion content corresponding to the input sentence obtained by the expansion unit and based on the history of the user arbitrarily selected by the user or by the user. By referring to the history of a group of selected users optionally, presents a candidate extended contents corresponding to the input sentence obtained to the user, document retrieval system according to claim.