JP2005208825A - Data search system, data search method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data search system capable of searching for a desired object based on an attribute related to the object without knowing the name of the object being searched for.
An ambiguous search database 105 stores a plurality of search data including a phrase indicating a name of an object to be searched and a phrase indicating an attribute related to the object. The search query input unit 107 inputs a search sentence in a natural language that does not include a word indicating the name of an object. The search query language analysis unit 108 performs natural language analysis on the search sentence, and obtains a natural language analysis result including identification of at least one of the phrases included in the search sentence that can be the attribute. The search unit 109 searches the search database based on the natural language analysis result. The search result output unit 110 outputs the search result.
[Selection] Figure 1

Description

  The present invention relates to a data search system, a data search method, and a program for searching for information based on a search sentence in a natural language.

  In recent years, convenience stores and large supermarkets are increasing. It deals with a wide variety of products and services, which are very convenient for consumers, but do not know where in the store what they need or where they can get the services they need. On the contrary, it may feel inconvenient. The point that it may be difficult to find a desired product is the same in other retail stores such as bookstores and CD shops. In addition, recently, homes and offices are full of things, and similar problems can occur.

  Therefore, for example, a search device may be installed in a retail store. For example, when a customer purchases a desired book or CD at a bookstore or CD shop and does not know the location of the desired book or CD in the store, the customer can use the search device to know the name and use the search device. The location of a desired book or CD, the presence or absence of inventory, etc. can be confirmed based on the information, and there may be a case where a sales floor for similar products can be searched from a store map or the like.

  However, when the name of a desired object cannot be remembered or the name of an object having a desired attribute (for example, function, form, content, etc.) is not known, a conventional search system that searches based on the name of an object is It was very difficult to use.

  Of course, you can ask the store clerk to explain the contents, functions, forms, etc., but you can't ask them if there are enough store clerk. You may even need to line up at the cash register, and the clerk may not be able to find it even though it actually exists in the store because it does not necessarily cover all the knowledge.

  Therefore, it is convenient if there is a system that can be searched based on attribute information such as contents, functions, and forms even if the names of goods and services are not known, but such a system did not exist.

Note that Patent Document 1 discloses a system for searching for information on a work of art based on preference / sense information. However, this system is a system that presents information on artworks that meet the user's preference when answering questions prepared in advance. It's not like finding something based on information other than your name.
JP 2002-297658 A

  Conventionally, it is difficult to search for an object when the name of the object being searched for is unknown.

  The present invention has been made in consideration of the above circumstances, and even if the name of the object that the user is looking for is not known, the data search that can search for the desired object based on the attribute related to the object It is an object to provide a system, an information search method, and a program.

  A data search system according to the present invention includes a search database for storing a plurality of search data including a phrase indicating a name of an object or service to be searched and a phrase indicating an attribute related to the object, and a search sentence in a natural language And a natural language analysis result including at least identification of a phrase that indicates the attribute among the phrases included in the search sentence by performing a natural language analysis on the search sentence It comprises analysis means, search means for searching the search database based on the natural language analysis result, and output means for outputting the search result.

The present invention relating to the apparatus is also established as an invention relating to a method, and the present invention relating to a method is also established as an invention relating to an apparatus.
Further, the present invention relating to an apparatus or a method has a function for causing a computer to execute means corresponding to the invention (or for causing a computer to function as means corresponding to the invention, or for a computer to have a function corresponding to the invention. It can also be realized as a program (for realizing the program), and can also be realized as a computer-readable recording medium on which the program is recorded.

  According to the present invention, it is possible to search for a desired object or service based on attributes related to the object or service without knowing the name of the object or service that the user is looking for.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  In this embodiment, the name of an object or service to be searched is associated with information indicating the attribute related to the object or service, information related to the attribute, or the like in a database, and the name of the object or service is known. Even if not, the user can search for a desired product or service based on attributes related to the product or service. This attribute includes an operation attribute (for example, an action for an object to be searched, an action for an object to be searched, an action of an object related to an object to be searched), a material attribute (for example, an object to be searched for) Material or material of the object to be searched, material or material of the object related to the object to be searched), environmental attributes (for example, the place where the object or service to be searched exists, what is the object or service to be searched Various places such as a place where an object or a service to be searched, a place where an object or a service to be searched is done, a place related to an object to be searched or an object related to a service), etc. In the specific example used below, an operation attribute is described as an example of this attribute.

  In each of the following embodiments, an example will be described, but a service can be similarly implemented.

(First embodiment)
FIG. 1 shows a configuration example of a data search system according to the first embodiment of the present invention.

  As shown in FIG. 1, this data search system includes a database registration content input unit (DB registration content input unit) 101, a database construction control unit (DB construction control unit) 102, a related document search unit 103, an ambiguous search database construction language. Analysis unit (fuzzy search DB construction language analysis unit) 104, fuzzy search database (fuzzy search DB) 105, search control unit 106, search query input unit 107, search query language analysis unit 108, search unit 109, search result output Unit 110 and an information database (information DB) 111.

  First, parts related to DB registration will be described.

  In the DB registration content input unit 101, a database registrant (DB registrant) who performs database registration for this data search system can register information indicating what information he / she wants to register. Note that this “what do you want to register information about” target is called “registration target entry”.

  The DB registrant can register information related to the registration target entry (related information) in the DB registration content input unit 101 together with the registration target entry. The data format of the related information is, for example, a tree structure of the result of sentence semantic analysis in natural language, but is not limited thereto. There is no particular limitation on the method of inputting related information. For example, the related information may be input as a document, or alternatively or in addition, a format for writing only necessary items is presented, and one of keyboard input, voice input, or file input is presented to the DB registrant. You may input by the method of having you input by two or more methods.

  In addition, the DB registrant wants to request a document (related document) as related information from the Web or an existing database, or uses the Web or an existing database in addition to the registered related document. When it is desired to request addition of a related document, an acquisition request or an addition request may be input to the DB registration content input unit 101.

  In the following, a case where the related information related to the registration target entry is a document (hereinafter also referred to as a related document) will be described as a main example. Further, the case where the data format of the related document is a tree structure of the semantic analysis result of the sentence in the natural language is described as a main example.

  The DB construction control unit 102 constantly monitors the presence or absence of a registration target entry in the DB registration content input unit 101 or when a registration target entry is registered in the DB registration content input unit 101. In this case, the entry to be registered is acquired or registered.

  Next, the DB construction control unit 102 creates data for registering the acquired registration target entry in the fuzzy search DB 105.

  When there is a related document acquisition request or addition request from the DB registrant, the DB construction control unit 102 searches for the related document using the registration target entry as a query (using the related document search unit 103). A related document is acquired from the Web or an existing database.

  The DB construction control unit 102, when a related document related to the entry to be registered is registered by the DB registrant in the DB registration content input unit 101, or when the DB construction control unit 102 receives an acquisition request or addition request from the DB registrant. When a related document is retrieved and acquired from the Web or an existing database, it is confirmed whether or not language analysis is necessary for the related document.

  For example, when the data format of the fuzzy search DB 105 is a tree structure of sentence semantic analysis results in natural language, the data format of a document related to the entry to be registered is a tree structure of sentence semantic analysis results in natural language. In some cases, there is no need for language analysis. In this case, the related document is registered in the fuzzy search DB 105 in association with the registration target entry without performing language analysis. On the other hand, in the same case, if the data format of the document related to the entry to be registered is a format other than the tree structure of the semantic analysis result of the sentence in natural language, there is a need for language analysis. At this time, the document related to the entry to be registered is converted into a tree structure of the semantic analysis result (using the ambiguous search DB construction language analysis unit 104), and this conversion result is registered in the fuzzy search DB 105 in association with the entry to be registered. To do. Also, for example, in the same case, when data is registered using a format in which only necessary items are written in the DB registration content input unit 101, there is a need for language analysis. The data is made into a tree structure, and this is associated with the registration target entry and registered in the fuzzy search DB 105.

  In response to an instruction from the DB construction control unit 102, the related document search unit 103 searches for a document related to the registration target entry from the Web or an existing database using the registration target entry registered in the DB registration content input unit 101 as a query. ,collect.

  The ambiguous search DB construction language analysis unit 104 holds at least one of a morphological analysis unit, a syntax analysis unit, or a semantic analysis unit. In response to an instruction from the DB construction control unit 102, at least one analysis of morphological analysis, syntax analysis, or semantic analysis is performed.

  The fuzzy search DB 105 is a fuzzy search DB in which the DB construction control unit 102 stores a related document related to a registration target entry obtained by an input to the DB registration content input unit 101 or a search by the related document search unit 103, or the related document. This is a database that stores at least one of the results analyzed by the construction language analysis unit 104, and includes a magnetic disk (floppy disk, hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, etc. This is realized using a storage medium.

  Heretofore, several methods for associating related information with registration target entries and registering them in the fuzzy search DB 105 have been shown, but the data search system may be capable of all those methods. However, only some of those methods may be possible. For example, in the latter case, a configuration in which one or both of the related document search unit 103 and the fuzzy search DB construction language analysis unit 104 are not provided is also possible. Of course, as a method for registering the related information in the fuzzy search DB 105 in association with the registration target entry, a method other than the method described so far may be used.

  Next, parts related to the search will be described.

  The search control unit 106 can accept a search query from a search query input unit 107 from a user (generally a person who is different from a DB registrant) who performs a search using this database registration. Upon receiving a search query corresponding to a search request from the user, the search control unit 106 performs a search based on the search query (using the search unit 109). At that time, the search query received from the search query input unit 107 may be searched as it is as a search query. If it is determined that language analysis is necessary for the search query, the search query language analysis unit 108 is set. The analysis result may be used, and the analysis result may be searched as a search query. The determination as to whether or not language analysis is necessary for the search query is based on matching between the search query input by the user and the database to be searched based on the search query. For example, when the search query entered by the user is a morpheme, if the database to be searched is a collection of documents, the search query itself can be matched with the database, so language analysis is not necessary. On the other hand, when the database to be searched has a tree structure of the semantic analysis result of the natural language sentence, it is necessary to convert the search query from the user into the tree structure of the semantic analysis result. Further, the search control unit 106 outputs the search result returned by the search unit 109 to the user using the search result output unit 110.

  Using the search query input unit 107, the user can input a search query using at least one of voice input, text input, or file input. The input may be a unit such as a morpheme, a tree structure, a phrase, or a sentence in natural language, and / or a format for writing only necessary items is prepared in the search query input unit 107 as an interface, and a user inputs a keyboard or You may enable it to input using at least one method of voice input or file input.

  The search query language analysis unit 108 holds at least one of a morpheme analysis unit, a syntax analysis unit, or a semantic analysis unit. In response to an instruction from the search control unit 106, at least one of morphological analysis, syntactic analysis, and semantic analysis is performed.

  In response to an instruction from the search control unit 106, the search unit 109 searches the information DB 111 for the search query received from the search control unit 106, and returns the search result to the search control unit 106. If no information matching the search query is obtained by this search, the search result indicates that information matching the search query could not be obtained.

  The search result output unit 110 can output the search result received from the search control unit 106 to the user using at least one method of voice input, text input, image input, or file input.

  Note that a series of information search processing using the search control unit 106, the search query input unit 107, the search unit 109, the search result output unit 110, and the information DB 111 (see the portion surrounded by a broken line in FIG. 1) is, for example, a special process. It may be equivalent to a conventional question answer generation technique represented by Kaihei 7-93351, Japanese Patent Laid-Open No. 7-192020, or the like.

  When the search control unit 106 receives a search result indicating that information matching the search query was not obtained as a result of the search using the information DB 111, the search query language analysis is performed. In order to obtain an analysis result suitable for the search in the fuzzy search DB 105 using the unit 108, the search query can be further analyzed. For example, when the information registered in the fuzzy search DB 105 has a tree structure obtained as a result of the semantic analysis, the search query language analysis unit 108 performs the semantic analysis. A tree structure like The search control unit 106 receives this tree structure and performs a search using the search unit 109 as a new search query. The search unit 109 uses the search query as a query for matching nodes and links in the tree structure, and performs a search in the tree structure.

  The information DB 111 is configured from the Web or an existing database, which is a storage medium such as a magnetic disk (floppy disk, hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, etc. To be realized.

  Hereinafter, the processing procedure of the data search system of this embodiment will be described in detail.

  First, the processing procedure of the DB construction control unit 102 of the data search system of this embodiment will be described.

  FIG. 2 shows an example of a processing procedure example of the DB construction control unit 102.

  In step S1, registration contents are accepted. In step S2, it is checked whether there is registration data in the DB registration content input unit 101. If there is registration data, the process proceeds to step S3, and if there is no registration data, the process returns to step S1.

  In step S <b> 3, registration contents are acquired from the DB registration contents input unit 101.

  Next, in step S4, it is checked whether or not the related document needs to be searched. If the related document needs to be searched, the related document is searched in step S5 and then the process proceeds to step S6. Skip S5 and proceed to step S6.

  Next, in step S6, it is checked whether or not language analysis is necessary for the related document. If language analysis is necessary for the related document, language analysis is performed in step S7 and then the process proceeds to step S8. If analysis is unnecessary, step S7 is skipped and the process proceeds to step S8.

  In step S8, registration is performed in the fuzzy search DB 105, and the process ends.

  Next, the processing procedure of the search control unit 106 of the data search system of this embodiment will be described in detail.

  FIG. 3 shows an example of the processing procedure of the search control unit 106.

  In step S11, a search query is accepted. In step S12, it is determined whether or not language analysis is required for the search query. If language analysis is required for the search query, language analysis is performed on the search query in step S13, and then the process proceeds to step S14. If language analysis is unnecessary for the query, step S13 is skipped and the process proceeds to step S14.

  In step S14, a search is performed.

  Next, in step S15, it is determined whether or not an answer can be obtained as a result of the search. If the answer can be obtained as a result of the search, the process proceeds to step S19 to output the search result. The process is terminated. On the other hand, if an answer cannot be obtained as a result of the search, the process proceeds to step S16.

  In step S16, it is determined whether or not language analysis of the search query is necessary again. If language analysis of the search query is necessary again, language analysis is performed again on the search query in step S17, and then step S18. If the language analysis of the search query is not necessary again, step S17 is skipped and the process proceeds to step S18. Then, the fuzzy search DB 105 is searched in step S18, the search result is output in step S19, and the process ends.

  Hereinafter, the present embodiment will be described in detail using specific examples.

  First, an example of creating the fuzzy search DB 105 based on the processing procedure of FIG. 3 is shown. Here, the word “canister” is registered (step S1). “Canister” entry to be registered and its related documents, “Canister” information obtained from the website, “The special rubber packing and clip can be used to seal tightly, jam, sauce, pickles, pasta It is assumed that the document having the text format “and the like can be stored for a long period of time” and the information that it is not necessary to add the related document are registered (step S1). Here, the data format of the fuzzy search DB 105 is a tree structure obtained as a result of semantic analysis.

  The DB construction control unit 102 checks whether or not data is registered in the DB registration content input unit 101 (step S2), and if there is a registration from the user, acquires the registration content from the DB registration content input unit 101. (Step S3). Here, “registration target entry”, “related document”, “required search for related document”, and “format of related document”, which are data registered by the user, are acquired.

  Next, it is checked whether or not a related document needs to be searched (step S4). Here, since there is an input that it is unnecessary, only the registered related document is used (step S5 is skipped).

  Next, it is checked whether or not language analysis is necessary for the related document (step S6). We have obtained a document in text format that says, “You can use a special rubber packing and clip to seal tightly and preserve jams, sauces, pickles, pasta, etc. for a long time.” Here, since the fuzzy search DB 105 uses the tree structure of the semantic analysis result as the data format, the information to be registered in the fuzzy search DB 105 is not an appropriate format as it is. Therefore, it is determined that language analysis is to be performed, and predetermined language analysis is performed (step S7). The data in the fuzzy search DB 105 can be data using a morphological analysis result or a syntax analysis result in addition to the tree structure of the semantic analysis result.

  FIG. 4 shows an example of a morphological analysis result for a related document “a special rubber packing and clip can be used to seal tightly and jam, sauce, pickles, pasta, etc. can be stored for a long period of time”. Here, part of speech information is given in addition to the headword, but in addition to this, utilization information and the like may be utilized for reading information, pronunciation information, and utilization words.

  FIG. 5 shows an example of the syntax analysis result, and FIG. 6 shows an example of the semantic analysis result.

  Here, the example of the semantic analysis result shown in FIG. 6 is used.

  The analysis result of FIG. 6 is registered in association with the registration target entry “canister” (step S8). For example, as shown in FIG. 7, “save” as a sub-variable is operated, “jam, source, pickles, pasta” is registered as “object”, and “long term” is registered as “adverb of time”. .

  Next, a specific example of search processing based on the processing procedure of FIG. 4 will be shown.

  The user can input a search query using at least one of voice, text, or file. Here, natural language input by voice is taken as an example.

  The user inputs, for example, “Which one stores pasta?” As a search query (step S11).

  In response to this input, first, a search is performed using the information DB 111 (step S14). As a result, for the search by the search query, the search control unit 106 receives that the information DB 111 has no answer to the search query and the result has not been obtained.

  Next, the search control unit 106 performs language analysis assuming that the search query needs to be searched in the fuzzy search DB 105 (step S17). In the processing procedure of FIG. 3, since the tree structure that is the result of semantic analysis is registered in the ambiguous search DB 105, here, semantic analysis is performed on the search query. The semantic analysis result is the same as in FIG.

  Using the analysis result as a new query, the fuzzy search DB 105 is searched (step S18). In the case of this search query, there is a tree in which there is a verb “save” in association with the question word “which”, and “pasta” as the object. Since the tree of the semantic analysis result with this “what” as the head that is the top node of the tree structure partially matches the data related to the canister in the registered data example of FIG. 7, “canister” is used as the search result. Output (step S19).

  In the above, when all of the natural language analysis results (in the example, the semantic analysis results) match all or part of the registration data, the registration data is acquired as satisfying the search query. However, for example, when a plurality of registered data matches, a plurality of registered data may be output as a search result, or ranking is performed according to a predetermined criterion, and only a specified number from the top is used as a search result. It may be output.

  Also, for example, when there is no registration data that matches the natural language analysis result, the condition is relaxed, and when a part of the natural language analysis result matches all or part of the registration data, the registration is performed. The data may satisfy the search query. If there is still no registration data that matches the natural language analysis result, the condition is further relaxed, and if a further part of the natural language analysis result matches all or part of the registration data, The registration data may satisfy the search query.

(Second Embodiment)
FIG. 8 shows a configuration example of a data search system according to the second embodiment of the present invention.

  As shown in FIG. 8, this data search system includes a database registration content input unit (DB registration content input unit) 101, a database construction control unit (DB construction control unit) 102, a related document search unit 103, an ambiguous search database construction language. Analysis unit (fuzzy search DB construction language analysis unit) 104, fuzzy search database (fuzzy search DB) 105, search control unit 106, search query input unit 107, search query language analysis unit 108, search unit 109, search result output Unit 110 and an information database (information DB) 111.

  The basic functions of each part are the same as those in the first embodiment. Below, it demonstrates centering on the point which is different from 1st Embodiment.

  Hereinafter, the processing procedure of the data search system of this embodiment will be described in detail.

  First, an example of the processing procedure of the DB construction control unit 102 of the data search system of this embodiment is the same as the processing procedure example of FIG.

  Next, the processing procedure of the search control unit 106 of the data search system of this embodiment will be described in detail.

  FIG. 9 shows an example of the processing procedure of the search control unit 106. Further, FIG. 10 shows a processing procedure a in step S30 of the processing procedure of FIG. The processing procedure b in step S42 of the processing procedure a in FIG. 10 is the same as the processing procedure in FIG.

  In step S21, a search query is accepted. In step S22, it is determined whether or not language analysis is necessary for the search query. If language analysis is necessary for the search query, language analysis is performed on the search query in step S23, and then the process proceeds to step S24. If language analysis is unnecessary for the query, step S23 is skipped and the process proceeds to step S24.

  In step S24, a search is performed.

  Next, in step S25, it is determined whether or not an answer has been obtained as a result of the search. If the answer has been obtained as a result of the search, the process proceeds to step S31 and the search result is output. The process is terminated. On the other hand, if no answer can be obtained as a result of the search, the process proceeds to step S26.

  In step S26, it is determined whether or not language analysis of the search query is necessary again. If language analysis of the search query is necessary again, language analysis is performed again on the search query in step S27, and then step S28 is performed. If the language analysis of the search query is not necessary again, step S27 is skipped and the process proceeds to step S28.

  In step S28, the fuzzy search DB 105 is searched.

  Next, in step S29, it is determined whether or not an answer with sufficient information can be obtained in response to the query from the user. If an answer with sufficient information can be obtained, the search result is output in step S31. To end the process.

  On the other hand, if it is determined in step S29 that an answer with sufficient information cannot be obtained, processing procedure a in step S30 is performed. That is, the registration target entry is registered in the DB registration content input unit 101 in step S41, the processing procedure b, that is, the processing procedure of FIG. 2 is executed in step S42, and the ambiguous search DB 105 is searched in step S43. In step S31, the search result is output and the process is terminated.

  Hereinafter, the present embodiment will be described in detail using specific examples.

  First, an example of search processing based on the processing procedure of FIG. 9 will be shown.

  The user can input a search query using at least one of voice, text, or file. Here, natural language text input using a keyboard is taken as an example.

  The user inputs, for example, “Is there anything that melts the hair of the bathroom pipe” as a search query (step S21).

  In response to this input, a search is first performed using the information DB 111 (step S24). As a result, for the search by the search query, the search control unit 106 receives that the information DB 111 has no answer to the search query and the result has not been obtained.

  Next, the search control unit 106 performs language analysis assuming that the search query needs to be searched in the fuzzy search DB 105 (step S27). Assuming that the tree structure that is the result of semantic analysis is registered in the fuzzy search DB 105, here, the search query is subjected to semantic analysis and the tree structure shown in FIG. 11 is output.

  Using this analysis result as a new query, a search in the fuzzy search DB 105 is performed (step S28). For example, if the tree structure of the registration data shown in FIG. 12 is in the fuzzy search DB 105, the action “melt” and the part “hair” as the object coincide with each other. is there. However, since there is no match for the information about the place “toilet”, an answer is returned that there is not enough information for the query.

  Therefore, the fuzzy search DB 105 is expanded according to the processing procedure a (step S30) shown in FIG.

  In the processing procedure a, “pipe cleaner” partially matched in the search for the fuzzy search DB 105 in step S28 is set as a registration target entry for expansion of the fuzzy search DB 105 (step S41).

  The DB construction control unit 102 checks whether there is DB registration content in the DB registration content input unit 101. Here, since there is a “pipe cleaner” as an entry to be registered, this is acquired (step S3).

  Next, it is checked whether or not a related document needs to be searched (step S4). Here, since the information regarding the related document and the necessity of searching for the related document is not registered, it is determined that the search is necessary.

  Using the Web, search for “pipe cleaner”, which is the entry to be registered, as a query (step S5). As a result, “drain pipe cleaner for pipes in bathrooms, washrooms, kitchens, etc. Get a document in text format.

  Here, since the fuzzy search DB 105 stores the tree structure of the semantic analysis result, it is determined that the text format is inappropriate as the format of information registered in the fuzzy search DB 105, and semantic analysis is performed (step S7). ). As a result, “bathroom”, “toilet”, “kitchen”, and “adhesive gel” as materials are extracted in association with “pipe cleaner”, and these are additionally registered in the fuzzy search DB 105 as indicated by the broken line portion in FIG. (Step S8).

  Thereafter, the fuzzy search DB 105 is searched (step S43), and the action "melt" matches "hair" as the object and "toilet" as the place, so "pipe cleaner" is output as the search result. (Step S31).

  The first and second embodiments can be implemented with various modifications. For example, the search result may be presented as an image, or the location of the search result may be presented using a map or the like. Further, the output may be set such that the robot or the like guides the user to a place where the search result exists, or the robot picks itself up. Further, the data search system may be installed in a store and used as a help desk. Further, a database such as synonyms and synonyms may be used to expand the database so that it can be used for searching.

Each of the above functions can be realized even if it is described as software and processed by a computer having an appropriate mechanism.
The present embodiment can also be implemented as a program for causing a computer to execute predetermined means, causing a computer to function as predetermined means, or causing a computer to realize predetermined functions. In addition, the present invention can be implemented as a computer-readable recording medium that records the program.

  Note that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of components disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

The figure which shows the structural example of the data search system which concerns on the 1st Embodiment of this invention. The flowchart which shows an example of the process sequence example of DB construction control part of the data search system of the embodiment The flowchart which shows an example of the process sequence of the search control part of the data search system of the embodiment Diagram showing an example of morphological analysis for related documents Diagram showing an example of parsing related documents Diagram showing an example of semantic analysis for related documents Figure showing an example of registered data The figure which shows the structural example of the data search system which concerns on the 2nd Embodiment of this invention. The flowchart which shows an example of the process sequence of the search control part of the data search system of the embodiment The flowchart which shows the processing procedure a of FIG. Diagram showing an example of semantic analysis for a search query Figure showing an example of registered data Figure showing an example of additional registration data

Explanation of symbols

  DESCRIPTION OF SYMBOLS 101 ... DB registration content input part, 102 ... DB construction control part, 103 ... Related document search part, 104 ... Language analysis part for fuzzy search database construction, 105 ... Fuzzy search database, 106 ... Search control part, 107 ... Search query input 108: Search query language analysis unit 109 ... Search unit 110 ... Search result output unit 111 ... Information database

Claims (7)

A search database for storing a plurality of search data including a phrase indicating a name of an object or service to be searched and a phrase indicating an attribute related to the object;
An input means for inputting a search sentence in natural language;
Analyzing means for performing a natural language analysis on the search sentence, and obtaining a natural language analysis result including identification of at least a phrase indicating the attribute among the phrases included in the search sentence;
Search means for searching the search database based on the natural language analysis result;
A data search system comprising output means for outputting the search result. The name is the name of the object to be searched,
The information on the attribute includes at least one of a phrase indicating an action related to the object, a word indicating a material related to the object, and a phrase indicating an environment related to the object. The data search system described in 1. The search data stored in the search database also includes a phrase indicating another attribute having a specific relationship with the attribute,
2. The data search system according to claim 1, wherein the natural language analysis result obtained by the analyzing unit includes a phrase indicating another attribute having a certain relationship with the attribute and specifying the certain relationship. . The search data stored in the search database has a data structure in the same format as the data structure of the natural language analysis result obtained by the analysis means,
The search means has a data structure including a data structure related to a natural language analysis result obtained by the analysis means for the search sentence among the search data stored in the search database, The data search system according to claim 1, wherein the data search system is acquired as satisfying a search sentence.   As a result of searching the search database by the search means, search data satisfying the search sentence is acquired because the search data stored in the search database does not include a phrase indicating a necessary attribute. In the case where it is impossible to do so, it is further provided with means for adding, to the search data, what may fall under a word indicating the required attribute based on information obtained from an external database different from the search database The data search system according to claim 1. Storing a plurality of search data including a word indicating a name of an object or service to be searched and a word indicating an attribute related to the object in a search database;
Steps for entering natural language search sentences;
Performing a natural language analysis on the search sentence, and obtaining a natural language analysis result including specifying at least a phrase indicating the attribute among the phrases included in the search sentence;
Searching the search database based on the natural language analysis result;
And a step for outputting the result of the search. In a program for causing a computer to function as a data search device,
The program is
Storing a plurality of search data including a word indicating a name of an object or service to be searched and a word indicating an attribute related to the object in a search database;
Steps for entering natural language search sentences;
Performing a natural language analysis on the search sentence, and obtaining a natural language analysis result including specifying at least a phrase indicating the attribute among the phrases included in the search sentence;
Searching the search database based on the natural language analysis result;
A program for causing a computer to execute the step for outputting the search result.

Images