JP5330049B2 - Search device, method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a retrieval technology allowing retrieval of dictionary data fit for an intention of a retriever. <P>SOLUTION: When a dictionary data management part 21 registers dictionary data into a dictionary DB 13, an index storage part 22 word-analyzes a certain kind of attribute included in the dictionary data, performs weighting in each word element according to the kind of the attribute, and registers an index. The dictionary data are retrieved by use of the index and the word element obtained by appropriately decomposing a keyword whose input is received by an input reception part 23. When the retrieved dictionary data are both pieces of the dictionary data corresponding to a class and the dictionary data corresponding to a property, a dictionary element ranking decision part 25 decides ranking according to correspondence relation between the class and the property. A retrieval result output part 26 transmits at least recommendation name included in each piece of the dictionary data to a retriever terminal 12 as a retrieval result in decided ranking order. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a search device, a method, and a program.

  Conventionally, technical documents exchanged between a supplier company that manufactures or sells a product and a user company that uses the product include a product catalog, a specification sheet, and a delivery specification. Unification of terms used in these documents, especially data items representing product specifications, is an urgent issue. Unification of terms and data items is required not only for data creation within the supplier company, but also between the user company and the supplier company from the viewpoint of the list of documents on the user company side. Recently, a supplier company and a user company have been working together to develop a dictionary (standard dictionary) that defines standard data items for each product category, so that data items can be standardized and unified. It has become to. One dictionary format is ISO13584 (commonly known as PLIB: Parts Library). PLIB describes a specific product specification according to a structure for defining a “property” that is a set of data items for expressing specification data of each product and a “class” that defines a product classification and a dictionary thereof. It is an international standard that defines a structure for exchanging “content”. In PLIB, “attributes” corresponding to detailed items for explaining what a data item means are standardized. A database that manages data according to PLIB can be regarded as one of hierarchical databases.

  International dictionaries conforming to PLIB include ISO13584-501 for measuring instruments and ISO13584-511 for bolts / nuts. In Japan, there are ECALS dictionaries of the Japan Electronics and Information Technology Industries Association and JeMarche dictionaries of the Japan Electrical Manufacturers' Association. Development of such dictionaries is also active all over the world.

  In the PLIB dictionary, “recommended name” is defined as one of “attributes” of “property”, and the term defined by this “recommended name” is also used in technical documents such as general product specification documents. It is considered desirable to be used. Even if an alias is used as a term, the concept itself preferably matches the “property” defined in the dictionary.

  Usually, a general domain engineer creates a technical document. For this reason, even an engineer who is not familiar with PLIB needs a function that can easily identify (search) the “property” and its “recommended name” that he / she envisioned in his / her head by inputting keywords. ing.

For example, there are the following two methods for searching dictionary data including “property” and “attribute” of the hierarchical database.
(1) A method in which several attributes are collectively searched for partial matching. For example, a method of outputting a property that partially matches a keyword input with respect to attributes such as “recommended name”, “alias”, and “short name” (see, for example, Patent Documents 1 and 2).
(2) A search method using a vector space method. For example, use full-text search, divide a document into words by N-gram or morphological analysis, extract feature vectors as typified by tf-idf, compare this with keyword vectors, and similar dictionary A method of retrieving data.

Further, although a little different from PLIB search, there are the following two methods for searching dictionary data in the field of general ontology search engines, for example.
(3) A method called OntoSelect that searches for partial matches on the ontologies on the Web for "class names" and "property names" and browses them. This is substantially equivalent to the method (1) described above.
(4) A method such as OntoRank or TermRank in Swoogle that ranks search results on the premise that the classes and properties referenced from more semantic Web contents are good ontologies.

JP 2004-177996 A JP 2004-178015 A

  However, in the methods (1) and (3), since individual dictionary data is a search target, for example, the following problems may occur. When a searcher inputs a keyword that crosses “class” and “property”, and the related dictionary data is searched and ranked in the order of high relevance, a property with the same “recommended name” is displayed on the dictionary. If it exists, there is a risk of presenting an incorrect ranking that does not match the searcher's intention.

  Further, when each dictionary data is regarded as a document and the dictionary data is searched by the method (2), for example, the following may be difficult. Since the definition in the dictionary data has a small number of characters, it is inaccurate to extract a feature vector therefrom, and the explicit relationship between “class” and “property” in the PLIB is reflected in the search result. Furthermore, it has been found that the method (4) has a problem that dictionary data having a higher ranking is not always applicable in the target area.

  The present invention has been made in view of the above, and an object of the present invention is to provide a search device, method, and program capable of searching dictionary data that matches a searcher's intention.

To solve the above problems, the present invention provides a searching apparatus is described by attribute a data dictionary data and each piece of property that a data is therefore described in the Attributes, for each class indicating the classification First storage means for storing dictionary data, and first analysis means for analyzing at least one first word element for each dictionary data by word-analyzing a certain kind of attribute value among the attributes of the dictionary data , Weighting means for weighting each first word element for each dictionary data according to the type of the attribute, the first word element, and a weight attached to the first word element Is stored in association with the dictionary data, first input receiving means for receiving keyword input, and the keyword as a word. Second analysis means for analyzing and obtaining at least one second word element; search means for searching for the dictionary data corresponding to the first word element matching the second word element; and the searched dictionary Ranking means for grouping the data into the same class or the same property, ranking the dictionary data for each group according to the weight assigned to the first word element, and the searched dictionary data into the class When there is both the corresponding dictionary data and the dictionary data corresponding to the property, a determination unit that determines the ranking of the dictionary data for each group according to the correspondence relationship between the class and the property, and Output means for outputting at least one of the attributes of the dictionary data in order.

Further, the present invention comprises a first storage means for storing the dictionary data described by a data dictionary data and each piece of the property attributes to be thus described Attributes, a data for each class indicating the classification A search method executed by a search device, wherein a word analysis is performed on a value of a certain kind of attribute among the attributes of the dictionary data to obtain at least one first word element for each dictionary data. A weighting step for weighting each first word element for each dictionary data according to the type of the attribute, the first word element, and a weight assigned to the first word element Is stored in the second storage means in association with the dictionary data, and a first keyword input is received. A force receiving step, a second analysis step of analyzing the keyword to obtain at least one second word element, and searching the dictionary data corresponding to the first word element that matches the second word element A ranking step of grouping the searched dictionary data into the same class or the same property, and ranking the dictionary data for each group according to a weight attached to the first word element; When the retrieved dictionary data includes both the dictionary data corresponding to the class and the dictionary data corresponding to the property, the ranking of the dictionary data for each group is determined according to the correspondence between the class and the property. Among the attributes of the dictionary data in the order of the confirmation step and the confirmed ranking, Both characterized in that it comprises an output step of outputting one.

  The present invention is also a program for causing a computer to execute the above-described method.

  According to the present invention, it is possible to search dictionary data that matches a searcher's intention.

  Exemplary embodiments of a search device, a method, and a program according to the present invention will be explained below in detail with reference to the accompanying drawings.

(1) Configuration First, the configuration of the search system according to the present embodiment will be described with reference to FIG. In the search system, the search device 10 is connected to the dictionary manager terminal 11 by wire or wireless, and is connected to the searcher terminal 12 by wire or wireless. The dictionary manager terminal 11 transmits dictionary data to be registered in a dictionary DB, which will be described later, to the search device 10. The searcher terminal 12 is a device in which a keyword is input in order for the searcher to search dictionary data, and transmits the input keyword to the search device 10.

  Next, the hardware configuration of the search device 10 will be described with reference to FIG. The search device 10 includes a control unit 101 such as a CPU (Central Processing Unit) that controls the entire device, and a storage unit such as a ROM (Read Only Memory) 104 and a RAM 105 (Random Access Memory) that store various data and various programs. And an external storage unit 107 such as an HDD (Hard Disk Drive) or a CD (Compact Disk) drive device for storing various data and various programs, and a bus 108 for connecting them, and a hardware using an ordinary computer The hardware configuration. The search device 10 includes a display unit 103 that displays information, an operation unit 102 such as a keyboard and a mouse that accepts user instruction inputs, and communication with the dictionary administrator terminal 11 and the searcher terminal 12 that are external devices. A communication I / F (interface) for controlling is connected to each other by wire or wirelessly.

  Next, in the hardware configuration described above, various functions that are realized when the control unit 101 of the search device 10 executes various programs stored in the ROM 104 or the external storage unit 107 will be specifically described. FIG. 3 is a diagram illustrating functional configurations of the search device 10. The search apparatus 10 includes a dictionary DB 13, a dictionary data management unit 21, an index storage unit 22, an input reception unit 23, a dictionary element ambiguous search unit 24, a dictionary element ranking determination unit 25, and a search result output unit 26. Have Among these, the dictionary data management unit 21, the index storage unit 22, the input reception unit 23, the dictionary element fuzzy search unit 24, the dictionary element ranking determination unit 25, and the search result output unit 26 are the control unit 101. Is generated on a storage unit such as the RAM 105 when the program is executed. The dictionary DB 13 is stored in the external storage unit 107, for example.

  The dictionary DB 13 stores dictionary data and an index table. The dictionary data is defined by a class indicating the classification and hierarchy of products and a property that characterizes the class. A property is described by at least one attribute. Dictionary data exists for each class and property. Specific examples of dictionary data and details of the index table will be described later. Products are provided to the market in order to meet the needs and desires of users. Not only tangible items but also services, experiences, events, people, places, assets, organizations, information, ideas Etc. are included in the product.

  FIG. 4 is a diagram illustrating a data structure of the dictionary DB 13. In the figure, it is shown that the dictionary DB 13 has a hierarchical structure centered on classification (class). In the hierarchical structure shown in the figure, the classification “semiconductor” is a root class, and “light emitting diode” and “light receiving element” exist as child classes. That is, it is shown that “semiconductor” is subdivided into two categories of “light emitting diode” and “light receiving element”. The hierarchical structure may have any number of layers. Symbols (in this example, C1 to C3) written in parentheses are codes that represent classes uniquely assigned in the dictionary DB 13, and correspond to identifiers called “class BSU codes” in the PLIB standard.

  As shown in FIG. 4, each class defining a hierarchical structure defines a unique property (a part surrounded by a dotted line in FIG. 4). In each class, the properties defined in the upper class of the class are inherited. For example, the properties of the class “light receiving element” are the properties defined in the class and the properties defined in the upper class “semiconductor”. In this way, classes and properties have a correspondence relationship. As in the case of a class, a code (“property BSU code” or simply “BSU”) (in this example, P0 to P3) representing a property uniquely assigned in the dictionary DB 13 is also assigned to the property. Hereinafter, both the code representing the class and the code representing the property are handled as the BSU code.

  FIG. 5 is a diagram illustrating dictionary data for each class. Each row corresponds to each dictionary data for each class. Each dictionary data includes a BSU code for identifying each class, “recommended name”, “alias”, “definition” as attributes, and a higher level of the class. Each includes a BSU code of a higher class for identifying the class.

  FIG. 6 is a diagram illustrating dictionary data for each property. Each row except the top row is a record corresponding to each dictionary data for each property. Each dictionary data includes a BSU code for identifying each property, a BSU code for identifying a class in which the property is defined (referred to as a definition class), and “recommended name”, “alias”, and “definition” as attributes. Each includes “unit”. Although not used in this embodiment, there are also attributes such as “short name”, “symbol”, “data type” of the property, and “source” indicating the source of the definition of the property itself. As one of the data models that define the structure of the dictionary such as the parent-child relationship of classes, the types of properties and their attributes, and the types of data types, ISO13584 (Parts Library: PLIB). In the present embodiment, description will be made using an application example where the dictionary conforms to the PLIB. However, the present invention may be applied to a database other than the PLIB.

  Returning to the description of FIG. The dictionary data management unit 21 manages the dictionary DB 13. Specifically, the dictionary data management unit 21 receives dictionary data transmitted from the dictionary manager terminal 11, registers it in the dictionary DB 13, or registers it in the dictionary DB 13. Controls access to dictionary data and corrects or deletes the dictionary data. When the dictionary data management unit 21 registers dictionary data in the dictionary DB 13, the index storage unit 22 performs word analysis on the value of a certain type of attribute included in the dictionary data, and each word (referred to as a word element). Each word element is weighted according to the type of attribute, and is registered in the index table as an index for each word element of each class and each property. Here, a certain kind of attribute is referred to as “recommended name”, “alias”, or “definition”.

  FIG. 7 is a diagram illustrating an index. As shown in FIG. 7, for each class and each property shown in FIGS. 5 and 6, the character strings that are the values of “recommended name”, “alias”, and “definition” that are attributes are respectively decomposed. An index for each word element is illustrated in each line. The index shown in each line is C / P for identifying whether it is a class or a property by C or P, the BSU code of the class or property, the word element, and the attribute name of the attribute decomposed into the word element. And weight values. Each index is associated with dictionary data of a class or property by a BSU code.

  Returning to the description of FIG. The input receiving unit 23 receives a keyword input and transmitted at the searcher terminal 12, analyzes the keyword, and decomposes it into each word element as appropriate. The dictionary element ambiguous search unit 24 searches the dictionary data using the word element obtained by appropriately decomposing the keyword received by the input receiving unit 23 and the index registered in the index table, and ranks the dictionary data. . Specifically, the dictionary element ambiguity search unit 24 searches the index table for an index record including a word element that matches the word element appropriately decomposed from the keyword, and finds the search result for the same class or the same property. The score of each dictionary data is calculated according to the weighting, and the dictionary data is ranked in the order of the score. The dictionary data corresponding to the BSU code included in the searched index becomes the searched dictionary data. If the dictionary data searched by the dictionary element fuzzy search unit 24 includes both dictionary data corresponding to a class and dictionary data corresponding to a property, the dictionary element ranking determination unit 25 ranks according to the correspondence relationship between the class and the property. Confirm. The search result output unit 26 transmits attribute values including at least “recommended name” included in each dictionary data to the searcher terminal 12 as a search result in the order of ranking determined by the dictionary element ranking determination unit 25.

(2) Operation Next, a procedure of processing performed by the search device 10 according to the present embodiment will be described. First, a registration process procedure in which the search device 10 registers dictionary data in the dictionary DB 13 and registers an index in the index table will be described. Upon receiving the dictionary data transmitted from the dictionary manager terminal 11, the search device 10 registers the dictionary data in the dictionary DB 13 by the function of the dictionary data management unit 21. Here, it is assumed that dictionary data to be registered is shown in FIGS. At this time, the search device 10 registers the index of the dictionary data in the index table by the function of the index storage unit 22. FIG. 8 is a flowchart showing a processing procedure for registering an index of dictionary data in the index table. The search device 10 stores the dictionary data to be registered shown in FIGS. 5 to 6 as dicElem (i, j) in a storage unit such as the RAM 105 (step S1). Here, i represents a dictionary data record corresponding to each row except the top row, and j represents an attribute corresponding to each column. As described above, the attribute types are “recommended name”, “definition”, and “alias” here, and the values of j are “1”, “2”, and “3” in order from the left. Next, for all dictionary data of dicElem (i, j) (step S2), the search device 10 sets the character strings represented as attribute values to words for all the target attributes (step S3). Analysis is performed (step S4). For example, in FIG. 5, for the attribute where i is “2” and j is “1”, that is, for the attribute “recommended name” in the class “C2” in the second row in FIG. A diode. In this case, the search device 10 decomposes this into “light emission” and “diode” by word analysis. Then, the search device 10 weights all the words (word elements) thus decomposed according to the type of the attribute, and stores the weight value (w (j)) in the index table together with the BSU code. Register (step S5).

In the present embodiment, the following values are used for the types of attributes as weights.
"Recommended name": 1.0
"Alias": 0.8
"Definition": 0.2

  As a result, for the attribute “recommended name” in the class “C2” on the second line in FIG. 5, the indexes Id1 to Id2 shown in FIG. 7 are registered in the index table. The search device 10 repeats the above processing for all the target attribute included in all the dictionary data to be registered. For the dictionary data shown in FIGS. 5 to 6, all the indexes shown in FIG. 7 are registered in the index table.

  Next, a search processing procedure for searching dictionary data based on a keyword input by a searcher will be described with reference to FIGS. When the search device 10 receives a keyword input and transmitted from the searcher terminal 12 by the function of the input receiving unit 23, the search device 10 analyzes the keyword and breaks it down into words (step S10). The word analysis method here is the same as the word analysis method performed in step S4. For example, it is assumed here that “diode rise time” is input as a keyword. At this time, in step S10, the search device 10 breaks it down into the words “diode”, “rising”, and “time”. Subsequently, the search device 10 performs the following processing by using the function of the dictionary element ambiguous search unit 24 using each decomposed word (word element). First, the search device 10 searches the index table for an index including a word element that matches each word element decomposed from the keyword (step S11). Then, the search device 10 groups the search results by the same class or the same property, and calculates the score of the dictionary data according to the weighting (step S12). In the present embodiment, the index and the dictionary data are associated with each other by the BSU code. For this reason, grouping is performed for each BSU code. In addition, the score is obtained by simply adding the weight values for each BSU code.

  In the example of each word element of “diode”, “rising”, and “time”, in step S12, indexes Id2 to Id9 shown in FIG. 7 are obtained as a result of the search. FIG. 11 is a diagram showing scores of dictionary data calculated by grouping the search results for each BSU code and weighting. In FIG. 11, the BSU code for identifying each dictionary data and the calculated score are shown for each dictionary data. The search result as above is the same as the result of the simple fuzzy search. If you refer only to this result, the score of property “P1” (“rise time” in class “C3”) is the highest. The result is that the dictionary data of the property “P1” is plausible as a search result.

Subsequently, when the dictionary data ranked by the function of the dictionary element ranking determination unit 25 includes both dictionary data corresponding to the class and dictionary data corresponding to the property, the search device 10 has a correspondence relationship between the class and the property. The ranking is determined according to (step S13). Specifically, the search device 10 adds the scores for each property and each class for the class and the properties defined in the class, and determines the order of the total score as a ranking. Here, the details of the procedure for determining the ranking will be described with reference to FIG. The class (definition class) in which the property Pi is defined is expressed as Pi.def_class, and the score of the dictionary data corresponding to the property Pi is expressed as Point ^ (Pi). Here, the definition class of each property is as follows.
P0.def_class = C1
P1.def_class = C3
P2.def_class = C2
P3.def_class = C2

The search device 10 adds the score of the definition class Pi.def_class to the score of the property Pi, and calculates the total score of the property Pi (step S20). Each total score is as follows, for example.
Point ^ (P0) = 0 + 0 = 0
Point ^ (P1) = 2.8 + 0 = 2.8
Point ^ (P2) = 2.6 + 1.0 = 3.6
Point ^ (P3) = 0 + 1.0 = 1.0

  Thereafter, the search device 10 ranks the dictionary data in the order of the total score (step S21). In the above example, the ranking order of properties is “P2> P1> P3> P0”.

  Then, the search device 10 does not change the score of the class itself, and ranks the dictionary data in the order of the score (step S22). In the example of FIG. 6, the ranking order for classes is “C2> C3 = C1”.

  As a result, the “recommended names” of properties P1 and P2 are both “rise time” and the same, but considering the relationship between the property and the definition class, any of the properties having the same recommended name Can be determined whether or not the searcher's intention is met. Therefore, it is possible to search for appropriate dictionary data.

  Returning to the description of FIG. Finally, the search device 10 transmits the value of the attribute including at least “recommended name” as each dictionary data to the searcher terminal 12 as a search result in the order of ranking determined in step S13 by the function of the search result output unit 26. (Step S14). FIG. 12 is a diagram illustrating search results. In FIG. 12, a BSU code for identifying dictionary data and a “recommended name” are shown as search results.

  As described above, word elements are extracted from certain types of attributes in the class dictionary data and the property dictionary data, and each word element is weighted according to the attribute type. As a result, with the ambiguity in the dictionary data, a search that takes into account the correspondence defined between the dictionary data, which is a feature of PLIB, for example, the relationship between the property and the class in which the property is defined, is performed. Can be realized. That is, even when there is a property with the same recommended name in the dictionary data, and the searcher inputs a keyword common to the class and the property and performs a search using this keyword, the weight attached to the word element is taken into consideration. Thus, the dictionary data can be presented with a certain ranking that matches the searcher's intention. Therefore, even a person who is not familiar with PLIB can easily identify (search) a “property” or “recommended name” by inputting a keyword. Therefore, it is possible to support the description using the recommended name and concept defined in the standard dictionary when creating a technical document such as a product specification document. In this way, the terms described in the product specification document are unified with the recommended names defined in the standard dictionary, thereby reducing the meaning interpretation between the supply company that creates the specification document and the user company that uses the specification document. And the quality of the specification document can be improved.

[Modification]
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. Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements 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. Further, various modifications as exemplified below are possible.

<Modification 1>
In the embodiment described above, various programs executed by the search device 10 may be stored on a computer connected to a network such as the Internet and provided by being downloaded via the network. The various programs are recorded in a computer-readable recording medium such as a CD-ROM, a flexible disk (FD), a CD-R, and a DVD (Digital Versatile Disk) in a file in an installable or executable format. May be configured to be provided.

<Modification 2>
In the embodiment described above, the search target class may be input together with the keyword at the searcher terminal 12 and transmitted to the search device 10. When receiving the class input together with the keyword, the search device 10 performs the above-described search process using only the dictionary data of the class among the dictionary data stored in the dictionary DB 13 as a search target. By limiting the search target in this way, the search process can be performed more efficiently.

<Modification 3>
In the embodiment described above, the index storage unit 22 excludes words such as particles, adverbs, and adjectives from registration when registering the index for each word element of each class and each property in the index table. Or only unknown words may be registered.

<Modification 4>
In the embodiment described above, the method for calculating the score of the dictionary data according to the weighting is not limited to the above example, and the method for determining the ranking according to the correspondence between the class and the property is not limited to the above example. .

<Modification 5>
In the above-described embodiment, the searcher may specify a class to be searched. In this case, the input receiving unit 23 receives input of a class identifier that can identify a search target class in addition to the keyword via the searcher terminal 12. The class identifier is a name such as a BSU code or “recommended name”, for example. The dictionary element ambiguous search unit 24 uses the word element obtained by appropriately decomposing the keyword received by the input receiving unit 23 and the index registered in the index table, and the search target class received by the input receiving unit 23 The dictionary data is searched and the dictionary data is ranked.

  FIG. 13 is a flowchart showing a procedure of search processing for searching dictionary data based on a keyword input by a searcher according to this modification. Step S10 is the same as that in the above-described embodiment. In step S30, the search device 10 includes the upper class of the search target class, the lower class including itself, the property defined in the upper class, and the property defined in the lower class including itself. The dictionary data corresponding to each is searched (step S30). A set of BSU codes of dictionary data to be searched is set as a search target set E. For example, when “light emitting diode (C2)” is designated as the class to be searched, the higher class is C1. Since there is no lower class, the lower class including itself is C2. The property defined in the higher class is P0. The properties defined in the lower class including itself are P2 and P3. Therefore, the search target set is “E = {C1, C2, P0, P2, P3}”.

Then, the search device 10 searches the index table for an index that includes a word element that matches each word element decomposed from the keyword among indexes whose BSU codes are included in the search target set (step S31). For example, when “diode rise time” is input as a keyword, the word elements decomposed from the keyword are “diode”, “rise”, and “time”. As a result of the search using each word element, in this modification, the indexes Id2, Id7 to Id9 shown in FIG. 7 are obtained in step S31. Then, the search device 10 groups the search results into the same class and the same property as in the above embodiment, and calculates the score of the dictionary data according to the weighting (step S32). FIG. 14 is a diagram showing scores of dictionary data calculated by grouping the search results for each BSU code and weighting. Step S13 is the same as that of the above-mentioned embodiment. The definition class of each property is as follows.
P0.def_class = C1
P2.def_class = C2
P3.def_class = C2

In this search target set example, the search device 10 calculates the total score of the property Pi by adding the scores of the definition class Pi.def_class to the score of the property Pi. It becomes.
Point ^ (P0) = 0 + 0 = 0
Point ^ (P2) = 2.6 + 1.0 = 3.6
Point ^ (P3) = 0 + 1.0 = 1.0
The order of ranking of properties is “P2>P3> P0”. Compared with the example of the above-described embodiment, since the property P1 and the class C3 are not targeted here, the ranking is different from that of the above-described embodiment. The ranking order for classes is “C2> C3”. After determining the ranking in this manner, the search device 10 transmits the search result to the searcher terminal 12 in the same manner as in the above-described embodiment in step S14.

  As described above, by narrowing down the search target class, it is possible to improve the processing efficiency of the search and improve the quality of the specification document.

<Modification 6>
In the above-described embodiment, when dictionary data can be embedded in the specification document in addition to superimposing the term on the specification document to the recommended word, the searched dictionary data itself or its BSU code is specified. It may be embedded in a document. FIG. 15 is a diagram showing an example of tag insertion when custom definition XML is used in accordance with the Office Open XML format. In FIG. 15, the BSU code of the dictionary data is embedded with the tag TG1, and the “recommended name” is embedded with the tag TG2. FIG. 16 is a diagram illustrating an example of a predefined custom XML schema. According to such a configuration, as a search result in step S14, for example, as shown in FIG. 17, although the tags TG1 and TG2 are not displayed, information for embedding such information is sent to the searcher terminal 12. Can be displayed.

It is a figure which illustrates the composition of the search system concerning one embodiment. It is a figure which illustrates the hardware constitutions of the search device 10 concerning the embodiment. It is a figure which illustrates each functional structure of the search device 10 concerning the embodiment. It is a figure which illustrates the data structure of dictionary DB13 concerning the embodiment. It is a figure which illustrates the dictionary data for every class concerning the embodiment. It is a figure which illustrates the dictionary data for every property concerning the embodiment. It is a figure which illustrates the index concerning the embodiment. It is a flowchart which shows the procedure of the process which registers the index of the dictionary data concerning the embodiment in an index table. It is a flowchart which shows the procedure of the search process which searches dictionary data based on the keyword which the searcher input. Details of the procedure for determining the ranking according to the embodiment will be described with reference to FIG. It is a figure which shows the score of the dictionary data calculated by grouping the search result concerning the embodiment for every BSU code | cord | chord, and weighting. It is a figure which illustrates the final search result concerning the embodiment. It is a flowchart which shows the procedure of the search process which searches dictionary data based on the keyword which the searcher concerning the modification of the embodiment inputs. It is a figure which shows the score of the dictionary data calculated according to the weighting which grouped the search result concerning the modification for every BSU code. It is a figure which shows the example of tag insertion when using custom definition XML according to the Office Open XML format concerning one modification. It is a figure which illustrates the schema of the predefined custom XML concerning the modification. It is a figure which illustrates the search result concerning the modification.

10 Search Device 11 Dictionary Manager Terminal 12 Searcher Terminal 13 Dictionary DB
21 Dictionary data management unit 22 Index storage unit 23 Input reception unit 24 Dictionary element ambiguous search unit 25 Dictionary element ranking determination unit 26 Search result output unit 101 Control unit 102 Operation unit 103 Display unit 107 External storage unit 108 Bus

Claims (8)

First storage means for storing the dictionary data described by attribute a data for each class indicating the classification A data for each dictionary data and properties that are thus written in Attributes,,
First analysis means for analyzing the value of a certain kind of attribute among the attributes of the dictionary data and obtaining each first word element for each dictionary data;
Weighting means for performing weighting according to the type of the attribute for each first word element for each dictionary data;
Second storage means for storing the first word element and the weight assigned to the first word element in association with the dictionary data;
First input receiving means for receiving keyword input;
Second analysis means for analyzing the keyword to obtain a second word element;
Search means for searching the dictionary data corresponding to the first word element matching the second word element;
Ranking means for grouping the searched dictionary data for the same class or the same property, and ranking the dictionary data for each group according to a weight attached to the first word element;
When the retrieved dictionary data includes both the dictionary data corresponding to the class and the dictionary data corresponding to the property, the ranking of the dictionary data for each group is determined according to the correspondence between the class and the property. A confirmation means;
A search device comprising: output means for outputting at least one of the attributes of the dictionary data in the determined ranking order. In the case where the retrieved dictionary data includes both the dictionary data corresponding to a class and the dictionary data corresponding to a property, the determining means determines the group according to the relationship between the class and the property defined in the class. The search device according to claim 1, wherein the ranking of each dictionary data is determined. A second input receiving means for receiving an input of a class identifier capable of identifying a class to be searched;
The retrieval means retrieves the dictionary data corresponding to the first word element that corresponds to the second word element and corresponds to the class specified by the class identifier. Item 3. The search device according to Item 1 or 2. The dictionary data is hierarchically arranged,
The search according to any one of claims 1 to 3, wherein the first storage unit stores the dictionary data in which the property belonging to a higher class of the hierarchy is inherited by the lower class. apparatus. 5. The search device according to claim 1, wherein the output unit outputs at least a recommended name among the attributes of the dictionary data in the determined ranking order. 6. The first analysis means performs word analysis on at least one value of a recommended name, an alias, and a definition that are types of attributes in the dictionary data, and each of the dictionary data includes at least one first word element. The search device according to claim 1, wherein the search device is obtained. Executed by the search apparatus comprising a first memory means for storing the dictionary data described a data of a data for each class indicating a classification for each dictionary data and properties that are thus written in Attributes, by attributes A search method,
A first analysis step of analyzing the value of a certain type of attribute among the attributes of the dictionary data to obtain at least one first word element for each dictionary data;
A weighting step for weighting each first word element for each dictionary data according to the type of the attribute;
A storage control step of storing the first word element and the weight assigned to the first word element in the second storage means in association with the dictionary data;
A first input receiving step for receiving an input of a keyword;
A second analysis step of analyzing the keyword to obtain at least one second word element;
A search step of searching the dictionary data corresponding to the first word element that matches the second word element;
A ranking step of grouping the searched dictionary data into the same class or the same property and ranking the dictionary data for each group according to a weight attached to the first word element;
When the retrieved dictionary data includes both the dictionary data corresponding to the class and the dictionary data corresponding to the property, the ranking of the dictionary data for each group is determined according to the correspondence between the class and the property. A confirmation step;
And a outputting step of outputting at least one of the attributes of the dictionary data in the determined ranking order.   A program for causing a computer to execute the method according to claim 7.

Images