JP2010092357A - Facility-related information retrieval method and facility-related information retrieval system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a facility-related information retrieval method to select information based on retrieval conditions of various viewpoints, and a facility-related information retrieval system adapted to retrieval conditions using a free sentence. <P>SOLUTION: The system has: an information collection device 12 which collects facility-related information, a facility information database 14 which stores the collected and organized facility-related information, and an input/output terminal 16. The system comprises an information retrieval device 18 which extracts a facility and the facility-related information adapted to request input by a retrieval user from the facility information database 14 and transmits the information to the input/output terminal 16, and a facility information index 20 and a dictionary database 22 storing the information related to the facility when the information retrieval device 18 retrieves the facility information database 14. The information collection device 12 comprises: an information collection means 12a, a keyword extraction means 12b which extracts keywords of content and feature of the facility, and an information organizing means 12c which adds a facility ID and the type of the facility and associates the facility ID with each keyword. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a facility-related information search method for collecting various facility-related information published in a website on the Internet, a commercial magazine, and the like, and facility-related information that provides facility-related information suitable for a search user's request input The present invention relates to an information retrieval system.

  In recent years, on occasions such as shopping, eating, sightseeing, etc., a predetermined database is accessed from a personal computer through a network such as the Internet, and facilities (for example, department stores, restaurants, amusement parks, scenic spots, etc.) suitable for the purpose are searched. Things are generally done. In this type of information search, it is required to accurately extract facility information that meets the search user's request, and various forms of search devices or search methods have been proposed in response to this requirement.

  For example, as disclosed in Patent Document 1, storage means for storing facility identification information provided from an information provider terminal and content information including facilities, products, etc. and their values in a database as individual information; , An association means for associating the content information of the individual information with a preset basic classification to generate an identification code, and searching the database by a keyword input from a consumer terminal, and an individual having an identification code suitable for it There is an information collection / retrieval device provided with means for reading out information and supplying it to a display device of a consumer terminal. This is an individual data set that is a set of data in a specific format that systematically defines identification codes such as product codes, characteristic / value codes, behavior codes, etc., and that provides identification codes that match each collected facility information And search for facility information according to the code system. In addition, the search by the consumer proceeds in an interactive manner using the screen of the consumer terminal.

Further, as disclosed in Patent Document 2, the present invention relates to an information search device such as telephone directory information, an input / output device that inputs an inquiry sentence and outputs information that matches the input information, and at least search target information A database that stores information including name, occupation name, and product name, a word dictionary storage device that stores predetermined words in the database and attributes of each word in pairs, and words related to occupations that appear in the inquiry sentence An information retrieval device comprising a knowledge base storage device in which a conversion table for converting names and verbs is defined, and a control device, and capable of performing a database search by estimating occupation classification names from ambiguous occupation information in query sentences There is.
JP 11-328223 A Japanese Patent No. 2666927

  However, the information collection and retrieval apparatus of Patent Document 1 extracts facility information having an identification code corresponding to a keyword input by a search user (consumer), with facility information identified by a predetermined code system. It does not have a function to evaluate how well the information meets the search user's request. Therefore, if there is a lot of facility information extracted as conforming to the search condition, the facility information is listed and output regardless of whether the fitness level of the facility information is high or low. It was troublesome to find out.

  In addition, when searching for facilities or the like, search users often place importance on evaluations of facility users such as “famous” or “popular” and reputation information such as word-of-mouth. However, the information collection / retrieval device of Patent Document 1 does not take into account the facility user's evaluation and the like, and may not be able to extract facility information that meets the search user's request.

  Furthermore, since the information collection / retrieval apparatus of Patent Document 1 performs a search in an interactive format, the search user needs to clarify his / her request in advance. Therefore, the search user cannot easily search for information on various facilities, and is a search device that is difficult to use.

  On the other hand, the information search apparatus of Patent Document 2 does not have a function for evaluating the degree of suitability of output information with respect to the request of the search user, and finds information close to the request from a large number of output results as in Patent Document 1. Was troublesome.

  Even if the search user inputs a query sentence in free text format, only specific nouns (product names) and verbs are taken up as search conditions, and other words are not considered. Therefore, for example, when this information search device is used for searching for entertainment / tourism facilities, etc., an inquiry sentence including abstract phrases such as adjectives and adverbs such as “famous” and “popular” is inputted. However, it was not possible to extract suitable facility information.

  The present invention has been made in view of the above-described background art, and collects a lot of facility-related information and makes it possible to select information based on search conditions from various viewpoints. It is an object of the present invention to provide a facility-related information search system that can cope with a search condition in a free text format including words and that can be output preferentially from facility information having a high degree of suitability for a search user's request.

  The present invention relates to a facility-related information search method that collects information about a facility, creates a database, and makes a search possible by a search server, an information collecting step of collecting facility-related information having document information, and the information collecting step For each facility-related information collected in step 1, a facility ID that identifies a facility that is the main body of the facility-related information and a facility ID that indicates a facility type that indicates the facility classification, and a facility that is assigned the facility ID Among the related information, at least the facility information storage step for storing the facility related information including the facility information in which the individual document information, the facility ID and the facility type are associated with each other is included in the storage device of the search server, and included in the facility related information. A keyword extraction process that divides document information into words and extracts keywords representing the contents and features of the facility from the words, A score assigning step for assigning a score value indicating the degree of association for each corresponding facility ID, and an index information storing step for storing index information including the keyword, the facility ID, and the score value in the storage device of the search server. The score assigning step includes a document information dividing step for dividing the document information of each facility-related information into a predetermined number of unit sentences from the beginning to the back of the document information, and the keyword extracted in the keyword extracting step. Is recognized in each unit sentence, the position of the unit sentence where the keyword appears, and the number of modifiers depending on the keyword in each unit sentence. The extraction key for the corresponding facility ID is based on at least one of the position of the front and the number of modifiers. The score value calculation process that calculates the weighted score value for the word relevance to each unit sentence, and the keywords that appear in each unit sentence are collected in each document information including the unit sentence, and different units A facility-related information search method comprising: a score value adding step of adding a score value assigned to each unit sentence when there is a keyword appearing redundantly in a sentence and setting the score value of the keyword in the document information It is.

  In the score assigning step, each keyword in which the score value is assigned to each document information in the score value adding step is collected in each facility ID, and if there is a keyword that appears redundantly in different document information, Among the assigned score values, a score value selecting step that selects the score value having the highest degree of relevance and sets the score value of the keyword in the facility ID is provided.

  When the number of sentences constituting the document information is less than the number of divisions set in advance in the sentence information dividing step, the sentence information dividing step treats the document information from the beginning to the back as one unit sentence, and the score In the value calculation step, a uniform score value is assigned to the keyword appearing in the unit sentence, which is less relevant than when the number of sentences constituting the document information is equal to or greater than a preset number of divisions. Is.

  Further, the present invention is a facility search system that is configured by a computer system, extracts a facility suitable for a request input by a search user, and provides the facility and the facility related information, and collects facility related information including document information. An information collection device for organizing, a facility information database for storing each of the facility-related information collected and organized, an input / output terminal operated by a search user, the facility suitable for the search user's request input, and the facility-related An information search device that extracts information from the facility information database and sends the information to the input / output terminal, and a facility information index that stores information related to the facility that is referred to when the information search device searches the facility information database And the dictionary database, and the information collection device includes the document information. Information collecting means for collecting facility-related information, keyword extracting means for extracting a keyword expressing the contents and characteristics of the facility from the document information included in the collected facility-related information, and individual facility-related information. On the other hand, a facility ID for identifying a facility that is the subject of the facility-related information and a facility type indicating the classification of the facility are given, and the facility ID of the facility corresponding to the document information is assigned to each keyword. An information organizing unit for performing an association process, and the facility information database includes at least the document information and the individual facility IDs assigned to the document information among the information organized by the information organizing unit. And each facility information including the facility type is stored in association with each other, and the facility information index stores information organized by the information organizing means. Are stored in association with the keyword and the facility ID associated with the keyword, and in the dictionary database, feature words representing the contents and features of the facility are set in advance. A dictionary in which the facility ID associated with each feature word is defined, or a dictionary in which the facility type associated with each feature word is defined, or both. An input means for inputting a sentence; and an output means for outputting or displaying a search result sent from the information search apparatus, wherein the information search apparatus divides and analyzes the request sentence sent from the input / output terminal. Word dividing means for obtaining word information, a dictionary of the dictionary database and the facility information index with reference to a feature word or keyword that matches the word information Extract the facility ID or the facility type defined to be related to the feature word or keyword, and extract the facility information to which the acquired facility ID or facility type is assigned from the facility information database A facility-related information search system provided with search means for sending to the input / output terminal.

  The keyword stored in the facility information index is given a score value indicating the degree of association for each corresponding facility ID by the information organizing unit, and the search unit extracts the facility information extracted from the facility information database. For each facility ID, the score value is aggregated to calculate the degree of match indicating the suitability for the request sentence, and the output means prioritizes the facility information to which the facility ID having a high match degree is assigned. Are output or displayed.

  The score value given to the keyword by the information organizing means includes the number of times the keyword appears in each document information, the appearance position in the document information, and the keyword in the document information. The keyword is determined based on the number of modifiers depending on the number of occurrences, the number of appearances is large, the position of the sentence relatively appears at the front position of the sentence information, and the keyword having a large number of modifiers has a higher degree of association. The score value weighted as follows is given.

  The dictionary database is a verb facility type dictionary in which feature words made up of verbs representing actions of users of the facilities are preset, and the facility types related to each feature word are individually defined by the information organizing means. The search means refers to at least the verb facility type dictionary and acquires the corresponding facility type.

  The dictionary database includes a place name dictionary in which feature words including place names representing the addresses of the facilities and related areas are set in advance, and the facility IDs related to the feature words are individually defined by the information organizing unit. The search means refers to at least the place name dictionary and acquires the corresponding facility ID.

  The dictionary database is an external knowledge word dictionary in which feature words made up of nouns that the user is interested in regarding the facility are preset, and the facility ID related to each feature word is individually defined by the information organizing means. The search means refers to at least the external knowledge word dictionary and acquires the corresponding facility ID.

  In the dictionary database, feature words made up of modifiers expressing the user's evaluation of the facility are set in advance, and the facility ID related to each feature word is individually defined by the information organizing means. A dictionary is provided, and the search means refers to at least the previous external evaluation dictionary and acquires the corresponding facility ID.

  In the dictionary database, feature words composed of modifiers expressing the evaluation of users of the facility are set in advance, and the facility ID associated with the information processing unit is individually defined in each feature word, and And an external evaluation word dictionary to which an evaluation value indicating the degree of association with the facility ID is individually given by the information organizing unit, and the search unit refers to at least the inside of the external evaluation dictionary, and the corresponding facility ID For the facility information extracted from the facility information database, the score value and the evaluation value are aggregated for each facility ID, and a matching degree indicating suitability for the request sentence is calculated, and the output means Is to output or display the facility related information corresponding to the facility ID having a high degree of match with priority.

  The facility-related information search method according to the present invention collects a huge amount of facility-related information disclosed on a website and the like, assigns at least index information including a facility ID, a keyword, and a score value and stores the index information in a search server. Thus, information can be organized and stored so that information can be selected based on search conditions from various viewpoints.

  The facility information retrieval system of the present invention further comprises an information collection device that collects and organizes facility-related information by the facility-related information retrieval method described above, a facility information database, a facility information index, and further includes a place name / verb / modifier. Since a dictionary database in which feature words such as words and facilities are associated with each other is provided, it is possible to interpret a search user's request from various viewpoints and accurately extract facility information that meets the request.

  Furthermore, since the information search apparatus can cope with a search condition in a free sentence format including an abstract word / phrase, the search user can easily perform a search without preparing in advance. In addition, the information search device weights each facility related information using an evaluation value indicating the evaluation of the facility user or the like, a score value indicating the degree of association between the facility and the keyword, and the degree of conformity to the request of the search user Since it is possible to preferentially output high facility information, it is very convenient for search users.

  Hereinafter, an embodiment of a facility-related information search system of the present invention will be described with reference to the drawings. The facility related information search system 10 is configured by a computer system, and, as shown in FIG. 1, an information collection device 12 that collects and organizes facility related information including document information, and a facility that stores the collected facility related information. An information database 14 and an input / output terminal 16 operated by a search user are provided. Furthermore, the facility corresponding to the request input of the search user and the facility related information are extracted from the facility information database 14 and sent to the input / output terminal 16, and the information search device 18 searches the facility information database 14. A facility information index 20 and a dictionary database 22 are provided to improve search efficiency and improve search accuracy.

  The information collecting device 12 is provided with information collecting means 12a capable of automatically collecting facility-related information including document information published on a website on the Internet, a commercial magazine, etc., or manually input by a system administrator. . Further, there is provided keyword extracting means 12b for dividing the document information included in the collected facility-related information into words using morphological analysis or the like, and extracting keywords expressing the contents and features of the facility from the words. . Furthermore, a facility ID that identifies the facility that is the subject of the facility-related information and a facility type that indicates the classification of the facility are given to each facility-related information, and each document before division is assigned to each keyword. An information organizing unit 12c that performs processing for associating the facility ID assigned to the information is provided.

  The facility information database 14 stores at least individual facility information composed of a combination of document information, a facility ID and a facility type corresponding to the document information, among the information organized by the information organizing unit 12c.

  The facility information index 20 is given a score value indicating the degree of association between the keyword and the facility ID to each keyword and the facility ID associated with the keyword among the information organized by the information organizing means 12c. Stored. The score value is determined based on the number of times the keyword appears in individual document information, the appearance position, and the number of modifiers that depend on the keyword. Then, a score value is given that the relevance degree is higher for a keyword that appears more frequently, appears at a position closer to the rear from the beginning of the document, and has a larger number of modifiers.

  In addition, the operation | movement applicable to one Embodiment of the facility relevant-information search method of this invention is performed by the information collection apparatus 12, the facility information database 14, and the facility information index 20. FIG. Details will be described later in the description of the operation of the facility information search system 10.

  As shown in FIG. 8, the dictionary database 22 includes a verb facility type dictionary 22a, a place name dictionary 22b, an external knowledge word dictionary 22c, and an external evaluation word dictionary 22d. In the verb facility type dictionary 22a, feature words composed of verbs expressing the behavior of the facility user are set in advance, and the facility type related to each feature word is defined by the information organizing unit 12c. In the place name dictionary 22b, feature words made up of place names that are nouns representing the address of the facility, a related area, and the like are set in advance, and the facility ID related to each feature word is individually defined by the information organizing means 12c. In the external knowledge word dictionary 22c, feature words including nouns that the user is interested in regarding the facility are set in advance, and facility IDs related to the feature words are individually defined by the information organizing unit 12c. In the external evaluation word dictionary 22d, feature words composed of adjectives expressing the evaluation of facility users or modifiers that are adjective verbs are set in advance, and the facility ID related by the information organizing unit 12c is assigned to each feature word. Evaluation values that are defined individually and indicate the degree of association for each facility ID are individually given by the information processing means 12c.

  The input / output terminal 16 includes input means 16a for inputting a request sentence in free text format by the search user, and output means 16b for outputting or displaying the search result sent from the information search device 18 to the search user. ing.

  The information search device 18 includes a word dividing unit 18a that divides a request sentence input to the input / output terminal 16 into word information using an analysis method such as morphological analysis, each dictionary and facility information index 20 included in the dictionary database 22. The feature word or keyword that matches the word information is extracted, the facility ID or facility type defined to be related to the corresponding feature word or keyword is acquired, and the acquired facility ID and facility type are given. A search means 18b is provided that extracts the facility information from the facility information database 14 and sends it to the output means 16b.

  The facility information search system 10 includes an input / output terminal 16 connected to a search server constituted by devices other than the input / output terminal 16 through a communication line such as the Internet. The search users can access the search server.

  Next, the operation of the facility related information search system 10 will be described. First, operations corresponding to the facility-related information search method 24 according to an embodiment of the present invention performed by the information collection device 12, the facility information database 14, and the facility information index 20 will be described with reference to the process diagram of FIG.

  First, in the information collection step S1, facility-related information published on a website on the Internet, a commercial magazine, etc., which includes at least document information, and other facility-related information to which images, diagrams, etc. are added is collected. , Capture on the computer. Next, in the facility ID etc. granting step S2, a facility ID for identifying the facility that is the subject of the information and a facility type indicating the facility classification are assigned to the collected individual facility related information. Then, in the facility information storage step S3, among the facility related information, at least the facility related information including the facility information in which the individual document information, the facility ID, and the facility type are associated is stored in the facility information database 14.

  For example, as shown in FIG. 3, if there are two pieces of facility-related information related to Westminster Abbey, a famous British church, the document information “the most beautiful in the UK…” The ID “1” is associated with the facility type “Temple” and stored in the facility information database 14 as the facility information 1. Similarly, for the second case, the document information “in medieval Gothic architecture ...”, sentence facility ID “1”, and facility type “Temple” are associated with each other as facility information 2 in the facility information database 14. Stored.

  On the other hand, after the information collecting step S1, the document information included in the facility related information is divided into words by morphological analysis or the like in the keyword extracting step S4 along with the processing of the facility ID etc. granting step S2, and Extract keywords that express content and features. Then, in the score assignment step S5, a score value indicating the degree of association is assigned to each extracted facility ID for the extracted keyword.

  Hereinafter, the score provision process S5 will be described in detail with reference to FIGS. The score providing step S5 includes a sentence information dividing step S51, a score value calculating step S52, a score value adding step S53, and a score value selecting step S54. First, in the document information dividing step S51, information consisting of one or a plurality of documents included in each facility related information is divided into unit sentences from the beginning to the back of the document information with one sentence as a minimum unit. For example, in the case of the example shown in FIG. 4, the document information 1 included in the first facility-related information related to Westminster Abbey is composed of three sentences. Here, for convenience of explanation, it is assumed that the document information is divided into three unit sentences, and in the document information 1, the first, second, and third sentences individually constitute one unit sentence.

  In this document information 1, keywords have already been extracted by the keyword extraction step S4. In the example shown in FIG. 4, in the first sentence, “recommended”, “world heritage”, “Westminster Abbey” are extracted as keywords, and “no”, “ha”, “wa” are important particles. It is not extracted because it is judged that it is not a proper word. Similarly, in the second sentence, “Westminster Abbey”, “Medieval”, “Beautiful”, and “Building” are extracted as keywords. “Westminster Abbey” overlaps with the first sentence, but is also treated as a keyword in the second sentence. The same applies to the third sentence.

  Next, in the score value calculation step S52, the number of times the keyword appears in each unit sentence, the position of the unit sentence in which the keyword appears, and the number of modifiers that depend on the keyword in each unit sentence And the degree of association between the keyword and the facility ID is calculated. In general, in document information that introduces facilities, there are many cases in which the document length is short, and the tendency that important words are used repeatedly may not apply. Therefore, attention is also paid to elements other than the number of occurrences of the keyword.

  First, it is determined that a keyword having a higher number of appearances is more important information and has a higher degree of relevance. However, in the example shown in FIG. 4, since the number of times each keyword appears in each sentence group is the same as each time, the value “1” is assigned to all the appearance times A here.

  Next, it is determined that the importance of the sentence group in which the keyword appears is closer to the head of the document information, and the importance is lower as it is closer to the back. This is based on an empirical rule that important information such as content that attracts the search user's interest and impressions of third parties tend to be written in unit sentences near the top. For example, in the example shown in FIG. 4, the weight B based on the keyword appearance position is “3” for the first sentence near the top, “2” for the second sentence in the middle, and “1” for the third sentence behind. "Is given.

  Further, it is determined that the importance is higher as the number of modifiers depending on the keyword is larger. For example, in the case of the example shown in FIG. 4, since the keyword “recommended” in the first sentence is not related to the modifier, “0” is used for the number C of modifiers and the keyword “Westminster Abbey”. Since there are two modifiers “recommended” and “world heritage”, “2” is assigned to the number C of modifiers. For other keywords, the number C of modifiers is given in the same manner.

  Then, by adding up the number of appearances A and the weight B, and further adding the number C of modifiers, a subtotal (A × B + C) that is a keyword score value for each unit sentence is calculated.

  Next, in the score value summarizing step S53, the keywords appearing in each unit sentence are collected in units of each document information, and if there are keywords that appear redundantly in different unit sentences, the score value assigned to each unit sentence is given. Are combined to obtain the score value of the keyword. For example, in the example shown in FIG. 4, the keyword “Westminster Abbey” appears in duplicate in the first, second, and third sentences, and subtotals (A × B + C) of “5”, “5”, and “3”, respectively. Is calculated. Therefore, the score value D (1) of “Westminster Abbey” based on the sentence information 1 is determined as “13” obtained by adding them. Since keywords other than “Westminster Abbey” do not overlap, the value of the subtotal (A × B + C) calculated for each unit sentence is directly used as the score value D based on the sentence information 1 without adding up the score values. (1) is determined.

  Further, the handling in the case where the object of the score giving step S5 is the second facility related information related to Westminster Abbey and the document information 2 is composed of one sentence as shown in FIG. 5 will be described. . Here, the weighting division number of the document information is set to 3, but since one sentence is a minimum unit, the document information 2 cannot be divided. Therefore, in the score value calculation step S52, the entire document information 2 is handled as one unit sentence.

  In the case of the text information 2, the number of appearances A of each keyword is given “1” that is the actual number of appearances as in the case of the document information 1. However, the weight B based on the appearance position of the keyword is uniformly assigned “0.9”. Based on an empirical rule that document information with a short length of text information is often poor quality information, it is determined that the importance is lower than document information with a long text length. Furthermore, the number C of modifiers is not reflected in the score value. Then, a subtotal (A × B) that is a score value of each keyword is calculated. In the next score value adding step S53, since the number of sentence groups of the sentence information 2 is one, the value of the subtotal (A × B) calculated for each keyword is directly used as the score value D based on the sentence information 2. (2) is determined.

  Next, in the score value selection step S54, the keywords whose score values are determined for each document information are collected for each facility ID, and if there are keywords that appear redundantly in different document information, the keywords are assigned to each document information. Among the score values, the score value having the highest degree of relevance is selected and set as the score value of the keyword. For example, in the case of the example shown in FIG. 6, the document IDs 1 and 2 are both given the facility ID “1” indicating Westminster Abbey. In the document information 1 and 2, the keywords “Westminster Abbey” and “building” are duplicated. If they overlap, the score value D (1) based on the document information 1 and the score value D (2) based on the document information 2 are compared, and the score value with the higher relevance level is selected and the score value of the facility ID “1” To decide. For keywords that do not overlap, the score values D (1) and D (2) are determined as the score values of the facility ID “1”.

  Then, after the score assigning step S5, in the index information storing step S6, the index information including the keyword, the facility ID, and the score value is stored in the facility information index 20 that is a database. As a result, the facility information index 20 stores index information relating to all facility IDs as shown in FIG. 7, for example.

  As described above, the information collection device 12, the facility information database 14, and the facility information index 20 are the information collection step S1, the facility ID etc. provision step S2, the facility information storage step S3, the keyword extraction step S4, and the score value provision step S5. And the facility related information retrieval method 24 comprising the index information storage step S6. According to the facility related information search method 24, a lot of facility related information is organized and stored so that information corresponding to search conditions from various viewpoints can be selected. Further, as in the facility-related information search system 10 described later, a search system that can accurately extract facility information suitable for a search user's request can be easily constructed, and furthermore, if a score value is used, It is also possible to provide in order from facility-related information with a high degree of relevance.

  Next, the operation of preparing the dictionary database 22 for improving the search efficiency of the information search device 18 and improving the accuracy of the search will be described with reference to the process diagram of FIG. In step S7, after the facility ID etc. granting step S2, a related facility is defined for each feature word set in each dictionary of the dictionary database 22, and the facility ID or facility type corresponding to the related facility is defined as each feature word. Store in association.

  In the verb facility type dictionary 22a of the dictionary database 22, feature words made up of verbs representing the operation of the facility user, such as “watch” and “play”, are set in advance (FIG. 8A). In general, in the facility introduction sentence created by the facility manager or the like, a verb whose operation is the subject of the facility user is not often used. On the other hand, verbs representing actions and actions of facility users are often used in request sentences input by search users. Therefore, even if a search by character string matching is performed using the verb input by the search user, the facility related information intended by the search user may not always be extracted. Therefore, by setting a feature word composed of a verb expressing the user's action and providing a verb facility type dictionary 22a defining a facility type related to the verb, it is possible to prevent the search omission and the like from occurring.

  Specifically, when the facility type is assigned to the new facility-related information in the facility ID etc. assigning step S2, and the facility type is new, it is set in the verb facility type dictionary 22a in step S7. An operation is performed to define that the new facility type corresponds to any of the feature words (verbs). In addition, when there is no appropriate feature word (verb), a new feature word may be set and an operation may be performed to define that it corresponds to the new facility type. For example, “watch” is a facility type corresponding to “art museum”, “museum”, etc., and “play” is a facility type corresponding to “amusement park”.

  In the place name dictionary 22b of the dictionary database 22, feature words made up of place names, which are nouns representing the address of the facility, the related area, etc., such as “London” and “△△ Fu XX City” are set in advance (see FIG. 8 (b)). Generally, there are many cases where the facility address does not include the address of the facility. Therefore, for example, even if the search user inputs the phrase “Italy” for the purpose of searching for a facility in Italy and performs a search by character string matching, information on a facility having an address in Italy can be extracted. In many cases, a number of Japanese Italian restaurants that do not meet the objectives are extracted. Therefore, by setting a feature word consisting of a place name representing a facility address, a related area, and the like, and providing a place name dictionary 22b that defines a facility ID related thereto, the above-mentioned search omission can be prevented.

  Specifically, when a facility ID is assigned to new facility-related information in the facility ID etc. granting step S2, if the facility ID is new, it is set in the place name dictionary 22b in step S7. An operation is performed to define that the new facility ID corresponds to any one of the feature words (place names). In addition, when there is no appropriate feature word (place name), a new feature word (place name) may be set, and an operation of defining that the new facility ID corresponds may be performed.

  In the external knowledge word dictionary 22c of the dictionary database 22, feature words made up of nouns that the user is interested in regarding the facility, such as “World Heritage”, “Drama writer”, and “Shakespeare”, are set in advance (FIG. 8 (c). )). In general, there are cases in which introductory text of a facility does not indicate that Shakespeare is a playwright even though there is a description that the facility is related to Shakespeare. Then, even if a search by character string matching using “drama writer” input by the search user is performed, there is a problem in that information on facilities that are related to Shakespeare cannot be extracted. Therefore, by setting a feature word composed of nouns that the user is interested in and providing an external knowledge word dictionary 22c that defines a facility ID related to the feature word, the above-mentioned search omission can be prevented. .

  Specifically, when a facility ID is assigned to new facility-related information in the facility ID etc. granting step S2, if the facility ID is new, it is set in the external knowledge word dictionary 22c in step S7. An operation is performed to define that the new facility ID corresponds to any of the feature words (nouns). In addition, when there is no appropriate feature word (noun) in the existing feature word, a new feature word may be set and an operation for defining that the new facility type corresponds may be performed. For example, the facility ID “1” of Westminster Abbey is defined to be related to “World Heritage”, “Historic Site”, and the like.

  In the external evaluation word dictionary 22d of the dictionary database 22, feature words made up of modifiers expressing the evaluations of facility users, such as “famous” and “fun”, are set in advance (FIG. 8D). . This feature word corresponds to an adjective or an adjective verb. In general, facility introductory texts created by facility managers and the like rarely use modifiers that express the reputation and spread of reviews and evaluations of the users of the facilities. In particular, there is no bad reputation. Therefore, in order to search for these appropriately, it is necessary to give the search system knowledge about the name and popularity of each facility. Therefore, it is possible to perform an appropriate search by setting a feature word composed of a modifier that expresses a user's evaluation and the like and providing an external evaluation dictionary 22d that defines a facility ID related thereto. Further, here, an evaluation value indicating the degree of association between the feature word (modifier) and the facility ID is also given.

  Specifically, when a facility ID is assigned to new facility-related information in the facility ID etc. granting step S2, if the facility ID is new, it is set in the external evaluation word dictionary 22d in step S7. An operation is performed to define that the new facility ID corresponds to one of the feature words (modifiers). In addition, when there is no appropriate feature word (modifier), a new feature word may be set, and an operation may be performed to define that the new facility type corresponds. For example, the facility ID “1” representing Westminster Abbey is associated with the modifier “famous”, and the degree of association is defined as “5”. On the other hand, the facility ID “2” representing the OO zoo is also related to the modifier “famous”, but is not as famous as Westminster Abbey, so the degree of association is defined as “2”. As described above, the evaluation value can be defined by any method. For example, the document information is analyzed by a predetermined method to automatically calculate the degree of popularity and popularity, or the administrator of the facility-related information search system 10 or the like is unique. There is a method of making a decision based on the survey and manually registering.

  In addition, by assigning an evaluation value to each facility ID, it is possible to perform an operation of providing facility-related information that meets the search user's request in descending order of relevance. This operation will be described in detail later.

  Next, an operation in which the search user operates the input / output terminal 16 to search for desired facility-related information will be described based on the flowchart of FIG. First, in step S <b> 10, the search user inputs a request sentence to the input unit 16 a of the input / output terminal 16. Here, for example, a request sentence in a free text form in which the purpose is expressed somewhat vaguely, such as “I want to see the architecture of a famous church in London” may be used.

  Next, in step S11, morphological analysis is performed by the word dividing means 18a, and the request sentence is divided into word information for each part of speech. At this time, synonyms and synonyms may be combined into word information using a thesaurus. In addition, when the request sentence contains a compound noun such as “National Park”, the thesaurus is searched for “National”, “Park”, and “National Park”, and the search results are aggregated to create a word for “National Park”. It may be information. That is, the method of extracting word information by the word dividing unit 18a is appropriately set so that the purpose of the search user can be properly recognized.

  Next, in step S12, the search means 18b refers to each dictionary in the dictionary database 22 and the facility information index 20, and the facility ID, evaluation value, and facility type corresponding to the word information obtained by the word dividing means 18 Get the score value. Hereinafter, the operation of the search means 18b will be described in detail based on the flowchart of FIG.

  If the word information is a verb, the search means 18b refers to the verb facility type dictionary 22a for a feature word (verb) that matches the word information, and obtains the corresponding facility type (step S121). For example, when the word information “view” is obtained by the word dividing means 18, the facility type such as “museum” or “museum” is acquired with reference to the verb facility type dictionary 22 a of FIG.

  When the word information is a place name, the search means 18b refers to the place name dictionary 22b, searches for a feature word (place name) that matches the word information, and obtains the corresponding facility ID (step S122). For example, when word information “London” is obtained by the word dividing means 18, a facility ID such as “1” is acquired with reference to the place name dictionary 22 b of FIG.

  When the word information is a noun, the search unit 18b refers to the external knowledge word dictionary 22c to determine whether there is a feature word (noun) that matches the word information (S123). If there is a matching feature word (noun), the search unit 18b acquires a facility ID corresponding to the word information (step S124). For example, when word information “world heritage” is obtained by the word dividing means 18, a facility ID such as “1” is acquired with reference to the external knowledge word dictionary 22 c of FIG.

  On the other hand, when there is no matching feature word (noun), the search unit 18b refers to the facility information index 20 to check a keyword that matches the word information, and acquires a corresponding facility ID and a corresponding score value ( Step S125). For example, when word information “Church” is obtained by the word dividing unit 18, the facility ID “1” and the score value “2” are acquired with reference to the facility information index 20 of FIG. 7.

  If the word information is a modifier, the search unit 18b refers to the external evaluation word dictionary 22d to determine whether there is a feature word (modifier) that matches the word information (S126). If there is a matching feature word (modifier), the search unit 18b acquires a facility ID corresponding to the word information and an evaluation value corresponding to the facility ID (step S127). For example, when the word information “famous” is obtained by the word dividing unit 18, the facility ID “1” and the evaluation value “5” are acquired with reference to the external knowledge word dictionary 22d in FIG. .

  On the other hand, when there is no matching feature word (modifier), the search means 18b refers to the facility information index 20 to check a keyword that matches the word information, and obtains the corresponding facility ID and the corresponding score value. (Step S128).

  Next, the search means 18b determines whether the facility ID or the facility type has been acquired through step S12 (step 13). When the facility ID and the facility type cannot be acquired from each dictionary and the facility information index 20 in step S12, the search unit 18b determines that “facility related information matching the request sentence is not found” The output means 16b outputs the information and the search operation is terminated (S14).

  On the other hand, when the facility ID or the facility type is acquired, the search unit 18b searches the facility information database 14 and extracts the facility information corresponding to the acquired facility ID and the facility type (step S15).

  Next, the search means 18b calculates the degree of match representing the suitability for the request sentence for each facility ID based on the score value and evaluation value given to the facility information extracted in step S15 (S16). For example, as shown in FIG. 11, when the facility ID “1” of Westminster Abbey is extracted, the facility ID “1” has an evaluation value “5” for the modifier “famous” and a keyword “building”. Is assigned a score value “4” and a score value “2” for the keyword “church”. Here, these are added together to calculate the degree of match “11” of the facility ID “1”. For the other facility IDs “7” and “12” extracted in step S15, the matching degree is calculated by the same method.

  And the search means 18b makes the facility information extracted by process S15 output from the output means 16b in an order with a high matching degree for every facility ID (S17). The output unit 16b provides the search result to the search user by printing on paper or by display display. As an output format, for example, a method of displaying on a display with a layout as shown in FIG. However, the system administrator can freely set which information is to be displayed among the document information, facility ID, match degree, etc. included in the facility related information, or the search user can select it. May be.

  As described above, the facility-related information search system 10 includes the information collection device 12, the facility information database 14, and the facility information index 20 that collect, organize, and store a large amount of facility-related information, and further includes place names / verbs / modifiers. Since the dictionary database 22 that associates the feature words such as and the like with the facility is provided, the request of the search user can be interpreted from various viewpoints, and the facility related information that meets the request can be accurately extracted.

  Furthermore, since the information search apparatus 18 can input a request sentence in a free sentence format including an abstract word / phrase, the search user can easily perform a search without preparing in advance. In addition, the information search device 18 weights each facility related information using an evaluation value indicating evaluation of the facility user, a score value indicating the degree of association between the facility and the keyword, etc., and conforms to the search user's request. It is possible to preferentially output high-level facility information, which is very convenient for search users.

  Note that the present invention is not limited to the above embodiment. For example, the number of types of dictionaries provided in the dictionary database 22 is increased / decreased depending on circumstances such that the search target is limited to a certain facility type. Can do.

  Also, feature words set in each dictionary, and score values and evaluation values assigned to feature words may lose their appropriateness with the passage of time, so maintenance (re-registration or correction) as necessary Need to do. In that case, the operation of correcting registration data such as a dictionary database and a facility information index is performed by operating a terminal device for a system manager (not shown). In this case, since index information such as a facility ID and a facility type is assigned to all the facility-related information accumulated, for example, processing such as correcting the information of a specific facility type at once is easy. Therefore, it is easy and easy to maintain.

  Further, for example, in a facility information retrieval system with a small number of facility-related information accumulated, the facility-related information extracted in step S15 is not necessarily provided with a function for calculating the degree of match using score values and evaluation values. It may be output in order.

  In addition to constructing a series of processing operations as in the above embodiment as a program, installing it in a server computer used as a facility-related information retrieval system, and executing it by a control means such as a CPU, the program is also transmitted via a network. May be distributed. In addition, the constructed program is stored in a portable storage medium such as a hard disk device, DVD-ROM, or CD-ROM connected to various computers used as a facility-related information retrieval system, and installed in the computer for execution. You may make it make it.

It is a block diagram which shows the structure of one Embodiment of the facility relevant-information search system of this invention. It is a processing process figure explaining one Embodiment of the operation | movement which the information collection device of this embodiment maintains a facility information database, a facility information index, and a dictionary database, and the facility related information search method of this invention. It is a figure which shows the example of the facility information database of this embodiment. It is a 1st figure explaining the score value stored in the facility information index of this embodiment. It is a 2nd figure explaining the score value stored in the facility information index of this embodiment. It is a 3rd figure explaining the score value stored in the facility information index of this embodiment. It is a 4th figure explaining the score value stored in the facility information index of this embodiment. It is a figure which shows the example of the dictionary with which the dictionary database of this embodiment is provided, Comprising: It is a verb facility classification dictionary (a), a place name dictionary (b), an external knowledge word dictionary (c), and an external evaluation word dictionary (d). It is a flowchart explaining the operation | movement which operates the input / output terminal of this embodiment, and outputs the searched facility information. It is a flowchart explaining the operation | movement in which the search means of this embodiment acquires the predetermined information applicable to word information with reference to a dictionary etc. It is a figure explaining the operation | movement which the search means of this embodiment calculates a matching degree. It is a figure explaining the example of the search result output from the output means of this embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Facility related information search system 12 Information collection apparatus 12a Information collection means 12b Keyword extraction means 12c Information arrangement means 14 Facility information database 16 Input / output terminal 18 Information search apparatus 18a Word division means 18b Search means 20 Facility information index 22 Dictionary database 22a Verb Facility type dictionary 22b Place name dictionary 22c External knowledge word dictionary 22d External evaluation word dictionary 24 Facility related information search method S1 Information collection step S2 Facility ID etc. addition step S3 Facility information storage step S4 Keyword extraction step S5 Score assignment step S51 Text information division step S52 Score value calculating step S53 Score value adding step S54 Score value selecting step S6 Index information storing step

Claims (11)

In a facility-related information search method that collects information about facilities, creates a database, and enables search by a search server,
An information collection process for collecting facility-related information with document information;
For each facility related information collected in the information collecting step, a facility ID etc. granting step for assigning a facility ID for identifying a facility that is a subject of the facility related information and a facility type indicating a classification of the facility,
Facility information storage that stores, in the storage device of the search server, facility related information including at least facility information associated with individual document information, facility ID, and facility type, among the facility related information to which the facility ID and the like are assigned Process,
A keyword extraction step of dividing the document information included in the facility-related information into words, and extracting keywords expressing the contents and features of the facility from the words;
A score giving step for giving a score value indicating a degree of association for each extracted facility keyword for each corresponding facility ID;
An index information storage step of storing index information including a keyword, a facility ID, and a score value in a storage device of a search server,
The score assigning step is a document information dividing step for dividing the document information of each facility-related information into a predetermined number of unit sentences from the beginning to the back of the document information;
Recognizes the number of times the keyword extracted in the keyword extraction step appears in each unit sentence, the position of the unit sentence in which the keyword appears, and the number of modifiers that depend on the keyword in each unit sentence And weighting that the degree of relevance of the extracted keyword to the corresponding facility ID is high with reference to at least one of the number of appearances and the position of the unit sentence in front and the number of modifiers A score value calculating step for calculating the score value for each unit sentence;
The keywords that appear in each unit sentence are compiled for each document information that includes the unit sentence. If there are keywords that appear in different unit sentences, the score values assigned to each unit sentence are added together. A facility-related information search method, comprising: a score value summation step that sets a score value of the keyword in document information.   In the score assigning step, each keyword in which the score value is assigned to each document information in the score value adding step is collected in each facility ID, and if there is a keyword that appears redundantly in different document information, 2. The score value selecting step of selecting a score value that has the highest degree of relevance from the assigned score values and setting it as the score value of the keyword in the facility ID is provided. Facility related information search method. When the number of sentences constituting the document information is less than the number of divisions set in advance in the sentence information dividing step,
The sentence information dividing step treats document information from the beginning to the back as one unit sentence,
In the score value calculating step, a uniform score value that is lower in relevance than a case where the number of sentences constituting the document information is equal to or more than a predetermined number of divisions with respect to the keyword that appears in the unit sentence. The facility related information search method according to claim 1, wherein the facility related information search method is provided. In a facility search system configured by a computer system, extracting a facility that matches a request input by a search user, and providing the facility and facility related information,
An information collection device that collects and organizes facility-related information including document information, a facility information database that stores each of the collected facility-related information, an input / output terminal operated by a search user, and the search user's An information search device that extracts the facility and the facility related information that are suitable for request input from the facility information database and sends them to the input / output terminal, and the facility that is referred to when the information search device searches the facility information database. A facility information index and a dictionary database storing information related to
In the information collection device, information collection means for collecting the facility-related information including the document information, and a keyword expressing the contents and characteristics of the facility are extracted from the document information included in the collected facility-related information. To each of the facility-related information, a facility ID for identifying a facility that is a subject of the facility-related information and a facility type indicating a classification of the facility, and for each of the keywords And an information organizing means for performing processing for associating the facility ID of the facility corresponding to the document information,
The facility information database includes individual facility information including at least the document information, the individual facility IDs assigned to the document information, and the facility type among the information organized by the information organizing unit. Stored in correspondence,
Of the information organized by the information organizing means, the facility information index stores the keyword and the facility ID associated with the keyword in association with each other.
In the dictionary database, feature words expressing the contents and features of the facility are preset, and the information organizing unit defines a dictionary in which the facility ID related to each feature word is defined, or is associated with each feature word. A dictionary in which the facility type is defined, or both are provided,
The input / output terminal comprises input means for inputting a request sentence in free text format, and output means for outputting or displaying a search result sent from the information search device,
The information retrieval device refers to a word division unit that obtains word information by dividing and analyzing the request sentence sent from the input / output terminal, matches the word information with reference to a dictionary of the dictionary database and the facility information index The facility ID or the facility type defined as related to the feature word or keyword is acquired, and the facility information to which the acquired facility ID or the facility type is assigned is acquired as the facility information. A facility-related information search system, comprising: search means for extracting from an information database and sending it to the input / output terminal. The keyword stored in the facility information index is given a score value indicating the degree of association for each corresponding facility ID by the information organizing means,
The search means, for the facility information extracted from the facility information database, calculates the degree of match indicating the suitability for the request sentence by counting the score values for each facility ID,
5. The facility related information search system according to claim 4, wherein the output means preferentially outputs or displays the facility information to which the facility ID having a high degree of match is assigned.   The score value given to the keyword by the information organizing means includes the number of times the keyword appears in each document information, the appearance position in the document information, and the keyword in the document information. The keyword is determined based on the number of modifiers depending on the number of occurrences, the number of appearances is large, the position of the sentence relatively appears at the front position of the sentence information, and the keyword having a large number of modifiers has a higher degree of association The facility related information search system according to claim 5, wherein the score value weighted as follows is assigned. The dictionary database is a verb facility type dictionary in which feature words made up of verbs representing actions of users of the facilities are preset, and the facility types related to each feature word are individually defined by the information organizing means. Prepared,
7. The facility-related information search system according to claim 4, 5 or 6, wherein the search means refers to at least the verb facility type dictionary and acquires the corresponding facility type. The dictionary database includes a place name dictionary in which feature words including place names representing the addresses of the facilities and related areas are set in advance, and the facility IDs related to the feature words are individually defined by the information organizing unit. ,
The facility related information search system according to claim 4, 5 or 6, wherein the search means refers to at least the place name dictionary and acquires the corresponding facility ID. The dictionary database is an external knowledge word dictionary in which feature words including nouns that the user is interested in regarding the facility are set in advance, and the facility IDs related to the feature words are individually defined by the information organizing unit. Prepared,
The facility related information search system according to claim 4, 5 or 6, wherein the search means refers to at least the external knowledge word dictionary and acquires the corresponding facility ID. In the dictionary database, feature words made up of modifiers expressing the user's evaluation of the facility are set in advance, and the facility ID related to each feature word is individually defined by the information organizing means. With a dictionary,
The facility related information search system according to claim 4, 5 or 6, wherein the search means refers to at least the previous external evaluation dictionary and acquires the corresponding facility ID. In the dictionary database, feature words composed of modifiers expressing the evaluation of users of the facility are set in advance, and the facility ID associated with the information processing unit is individually defined in each feature word, and And an external evaluation word dictionary to which an evaluation value indicating the degree of association with the facility ID is individually given by the information organizing means,
The search means refers to at least the inside of the external evaluation dictionary, acquires the corresponding facility ID, and obtains the score value and the evaluation value for each facility ID for the facility information extracted from the facility information database. Aggregate to calculate the degree of match that indicates the appropriateness of the request statement,
7. The facility related information search system according to claim 5, wherein the output means preferentially outputs or displays the facility related information corresponding to the facility ID having a high degree of match.

Images