JP5578137B2 - Search program, apparatus and method - Google Patents

Description

  The present technology relates to search technology.

  When a plurality of databases that store different types of data are search targets, when a user inputs a search query, the plurality of databases are searched based on the search query, and data that matches the search query is detected. As a conventional technique for searching a plurality of databases, there is known a technique in which a dedicated search screen is displayed on a database when the user selects one database from the plurality of databases. The selected database is searched with the keyword input on the search screen.

  However, in the above prior art, for each of a plurality of databases, the user inputs a search query on a search screen dedicated to the database. In addition, in order to increase the accuracy of the search, the user himself grasps the characteristics of the data in each database and considers what kind of keyword is suitable for which database. For this reason, it was not convenient for the user.

  On the other hand, as a conventional technique, a method in which a user inputs a common search query for all databases is also known. However, when different types of data are stored in each of a plurality of databases, a common search query includes keywords that are inappropriate for the type of data, resulting in a decrease in search accuracy. For this reason, the user may not be able to obtain desired data.

  In particular, in recent years, data relating to one entity is often distributed and stored in a plurality of databases. For example, when the entity is a product, a book, a store, etc., the basic data includes elements such as the product name and function of the product, the bibliographic items of the book, and the type and location of the store. To this basic data, additional data such as data including evaluation elements such as cheap, interesting and delicious and data including ranking elements such as popularity and price are added. These data may be distributed and stored in a plurality of databases. In particular, in Internet blogs and the like, data including evaluation elements is often added to basic data. In addition, data sharing is progressing within a company, and a plurality of people may add data to one entity. As described above, various types of data are added with respect to one entity, and cases in which this data is distributed and stored in a plurality of databases are increasing, and it is difficult to obtain desired data at the time of search. .

JP 2007-133505 A

  Therefore, the objective of this technique is providing the technique for improving a user's convenience, when searching a some database as one side surface.

  In this search method, (A) one or more query elements included in a first search query input by a user are extracted, and (B) a plurality of databases each storing different types of data for each query element. (C) Based on the degree of association, the presence / absence of association with each of the plurality of databases is determined for each query element, and (D) each of the plurality of databases is associated with the database. If there is a query element determined to be, a second search query including the query element is generated. (E) Based on the second search query, the second search query is selected from the plurality of databases. Includes processing to search the corresponding database.

  When searching a plurality of databases, user convenience can be improved.

FIG. 1A is a diagram for describing Example 1 of a database to be searched in the embodiment of the present technology. FIG. 1B is a diagram for explaining Example 2 of the database to be searched in the embodiment of the present technology. FIG.1 (c) is a figure for demonstrating the example 3 of the database used as the search object in embodiment of this technique. FIG. 2 is a functional block diagram of the search device. FIG. 3 is a diagram illustrating a processing flow of processing by the search device. FIG. 4 is a diagram illustrating a processing flow of processing of the extraction unit. FIG. 5 is a diagram illustrating a specific example of processing by the extraction unit. FIG. 6 is a diagram illustrating a first example of a processing flow of processing of the calculation unit and the determination unit. FIG. 7 is a diagram illustrating a second example of the processing flow of the calculation unit and the determination unit. FIGS. 8A to 8B are diagrams for explaining a specific example of the processing of the calculation unit and the determination unit. FIG. 9 is a diagram illustrating a first example of a processing flow of the processing for calculating the degree of association. FIG. 10 is a diagram illustrating a second example of the process flow of the process of calculating the degree of association. FIG. 11 is a diagram illustrating an example of a category table. FIG. 12 is a diagram illustrating a processing flow of processing by the query generation unit. FIGS. 13A and 13B are diagrams illustrating a specific example of processing by the query generation unit. FIG. 14A is a diagram illustrating a processing flow of processing by the search unit. FIG. 14B is a diagram illustrating a processing flow of processing for selecting a search method according to a category. FIG. 15 is a diagram illustrating a first example of a processing flow of processing performed by the set generation unit. FIG. 16 is a diagram illustrating a specific example of the processing flow example 1 of processing by the set generation unit. FIG. 17 is a diagram illustrating a second example of a processing flow of processing performed by the set generation unit. FIG. 18 is a diagram illustrating a specific example of the processing flow example 2 of processing by the set generation unit. FIG. 19 is a diagram illustrating an example of data representing the positional relationship of each region. FIG. 20 is a diagram illustrating a first example of a processing flow of processing performed by the subset generation unit. FIG. 21A to FIG. 21C are diagrams illustrating a specific example of the processing flow example 1 of processing by the subset generation unit. FIG. 22 is a diagram illustrating a second example of the process flow of the process performed by the subset generation unit. FIG. 23 is a diagram illustrating a second example of the process flow of the process performed by the subset generation unit. FIGS. 24A to 25C are diagrams illustrating a specific example of the processing flow example 2 of processing by the subset generation unit. FIG. 25 is a functional block diagram illustrating a search device according to another embodiment. FIG. 26 is a functional block diagram illustrating a search device according to another embodiment. FIG. 27 is a functional block diagram of a computer.

  First, an outline of the embodiment of the present technology will be described with reference to FIGS. 1 to 3. In this embodiment, there are a plurality of databases to be searched, and different types of data are stored. In the data stored in the database, a plurality of data blocks are associated with one entity, and these data blocks are distributed and stored in a plurality of databases. FIG. 1A to FIG. 1C are examples of databases to be searched in this embodiment.

  FIG. 1A is a diagram for explaining Example 1 of a database to be searched. In Example 1, an entity is associated with a plurality of data blocks by an identifier. The identifier is, for example, a book title, a store name, a literature reference number, or the like. By assigning this identifier to each data block, the entity and the data block are associated with each other. In FIG. 1A, the data block d1 to which the identifier e1 is assigned is associated with the entity E1, the data block d2 to which the identifier e2 is assigned is associated with the entity E2, and the data block d3 to which the identifier e3 is assigned is the entity Associated with E3. Each data block d1 to d3 is distributed and stored in a plurality of databases according to the type. For example, if the entity is a book, a data block in which bibliographic items are described, a data block in which evaluation is described, and the like are stored in different databases. Here, the identifier includes not only an identifier as a strict unique name such as a serial number for identifying each individual, but also a name that can specify the type and classification such as a product name.

  FIG. 1B is a diagram for explaining an example 2 of a database to be searched. In Example 2, an entity is associated with a plurality of data blocks by reference. For example, there is a main data block for an entity and an ancillary data block, and the ancillary data block has a link to the main data block. The main data block and the accompanying data block are distributed and stored in a plurality of databases. A typical example is a Web annotation. Web annotation is a technique for giving additional information to a Web page. For example, when a tag (memo) is electronically given to a Web page, the Web page is a main data block, and a memo with a link on the Web page is an associated data block. In FIG. 1B, for the entity E1, the data block d42 refers to the data block d41, for the entity E2, the data block d52 refers to the data block d51, and for the entity E3, the data block d62 refers to the data block d61. doing. The main data blocks d41, d51, d61 and the accompanying data blocks d42, d52, d62 are distributed and stored in different databases.

  FIG. 1C is a diagram for explaining Example 3 of the database to be searched. In Example 3, an entity and a plurality of data blocks are associated by dividing data. The data related to each entity is physically one data block, and a plurality of types of information are included therein. This data block is virtually separated for each type of information, and each of the separated parts is regarded as a virtual data block. The virtual data block is regarded as being distributed and stored in the virtual database. A typical example is an entity introduction site. The entity data block includes basic information and evaluation information. The basic information is, for example, information such as bibliographic items of books, store types, store names, and maps. The evaluation information is information such as description of evaluation and score for the entity. The basic information and the evaluation information are virtually separated, and each separated part is regarded as a virtual data block. The basic information database and the evaluation information database are considered to be virtually, and the virtual data blocks are assumed to be distributed and stored in the virtual database. In FIG. 1C, virtual data blocks d71, d72, d73 are included in the data block related to the entity E1, and virtual data blocks d81, d82 are included in the data block related to the entity E2. , Virtual data blocks d91 and d92 are included in the data block related to the entity E3. Note that the data blocks d71, d81, d91 and the data blocks d72, d73, d82, d92 are considered to be distributed and stored in a plurality of virtual databases according to their types.

  As shown in Examples 1 to 3, when a plurality of databases each storing different types of data are to be searched, if the user inputs a dedicated search query for each database, the convenience is lowered. On the other hand, if a search is performed based on a search query common to a plurality of databases, the search accuracy is low because query elements that do not match the data type of the database are included. In addition, the search result is output for each database, which is complicated for the user. Even if the user is seeking an entity that matches the search query, the search result is given as a sequence of data blocks that match the search query, so it does not necessarily match the information needs of the user. The technique of the present embodiment can solve such a problem.

  FIG. 2 is a functional block diagram of the search device 100 according to the present embodiment, and FIG. 3 is a diagram illustrating a processing flow of processing by the search device 100. The search device 100 can search a plurality of databases D1, D2, and D3. Different types of data are stored in each of the databases D1, D2, and D3. In the present embodiment, a plurality of data blocks are associated with one entity, and the plurality of data blocks are distributed and stored in the databases D1, D2, and D3. Examples of the database are as shown in FIGS. 1A, 1B, and 1C. In the present embodiment, the number of databases is three, but the number of databases is not limited to three. The search device 100 includes an extraction unit 1, a calculation unit 2, a determination unit 3, a search query generation unit 4, search units 51, 52, and 53, set generation units 61, 62, and 63, and a subset generation unit. 7, a query list storage unit 10, a query storage unit 11, a set storage unit 12, and a subset storage unit 13.

  Hereinafter, each part will be described. Note that each step in FIG. 3 is executed corresponding to each part, and therefore the description of each step is performed by associating each step with each part. The extraction unit 1 has a function of extracting one or a plurality of query elements qi included in the first search query Q1 input by the user (step S1). The calculation unit 2 has a function of calculating the degree of association f (qi, Dj) with each database Dj (D1, D2, D3) for each query element qi extracted by the extraction unit 1 (step S3). . The determination unit 3 determines whether each query element qi is related to each database Dj (D1, D2, D3) based on the degree of association f (qi, Dj) calculated by the calculation unit 2 ( Step S5) A function is provided. When there is a query element qi determined to be associated with the database Dj (D1, D2, D3) for each of the databases Dj (D1, D2, D3), the search query generation unit 4 sets the query element qi to A second search query Q2dj including the function is generated (step S7). The search units 51, 52, and 53 have a function of searching the database Dj corresponding to the second search query Q2dj from the databases D1, D2, and D3 based on the second search query Q2dj (step S9). The search unit 51 is in charge of searching the database D1, the search unit 52 is in charge of searching the database D2, and the search unit 53 is in charge of searching the database D3. When the data generation is detected from each of two or more databases among the databases D1, D2, and D3 in the processing by the search units 51, 52, and 53, the set generation units 61, 62, and 63 are associated with the data blocks. A function is provided for identifying entities and generating a set of entities (step S11). The set generation unit 61 processes the search result of the search unit 51, the set generation unit 62 processes the search result of the search unit 52, and the set generation unit 63 processes the search result of the search unit 53. The subset generation unit 7 has a function of generating a subset of entities from the set of entities specified by the set generation units 61, 62, and 63 based on a predetermined rule (step S13).

  The search device 100 is used as follows. The user inputs the first search query Q1. At this time, the user may input the first query Q1 as a search query common to the plurality of databases D1, D2, D3 without being aware of the type of data stored in the databases D1, D2, D3. The user does not need to input a search query for each of the databases D1, D2, and D3, and does not need to consider what kind of keyword is suitable for which database. Whatever query element is included in the first search query Q1, whether or not there is a relation is determined based on the degree of relation described later, so that the query element to be used is not limited and the first search The degree of freedom of query is high.

  When the first search query Q1 is input, the extraction unit 1 extracts one or a plurality of query elements qi from the first search query Q1. The calculation unit 2 calculates the degree of association f (qi, Dj) with the databases D1, D2, and D3 for each of the query elements qi. The determination unit 3 determines whether or not each query element qi is associated with the databases D1, D2, and D3 based on the degree of association f (qi, Dj). Whatever query element is included in the first search query Q1, the degree of association f (qi, Dj) is calculated each time, and the presence or absence of the relation is determined. Then, the search query generation unit 4 generates a second search query Q2dj for each of the plurality of databases Dj (D1, D2, D3). The second search query Q2dj includes a query element determined to be related to the database Dj. The search units 51, 52, and 53 search the database Dj corresponding to the second search query Q2dj based on the second search query Q2dj. The search units 51, 52, and 53 search for the responsible database Dj (D1, D2, D3) based on the second search query Q2dj suitable for the responsible database Dj (D1, D2, D3). , Search accuracy will be higher. When the search units 51, 52, and 53 detect data blocks from two or more databases, the set generation units 61, 62, and 63 identify entities associated with the data blocks, and generate a set of entities. . Then, the subset generation unit 7 generates a subset from the set of entities based on a predetermined rule. A list of entities included in this subset is output to the user. The list of entities may be a list of entity identifiers or a list of links to data about the entities. The entity subset is obtained by integrating the search results of the databases D1, D2, and D3 based on a predetermined rule. According to the present embodiment, the user can obtain an integrated search result instead of obtaining a search result for each database. Since the search results are given as a set of entities, it matches the user's information needs.

  Hereinafter, an example of each of the above-described units and steps will be described in detail.

  FIG. 4 is a diagram illustrating an example of a processing flow of the processing of the extraction unit 1, and corresponds to step S1 of FIG. In the present embodiment, the first search query Q1 is input by the user as natural language text data. The extraction unit 1 takes in the first search query Q1 (step S101), performs morphological analysis (step S103), and removes stop words (step S105). A stop word is a word that is excluded from words used for search, such as a function word. Next, the extraction unit 1 collects morphemes from a semantic viewpoint by chunking (step S107), and shapes a word form from a retrieval viewpoint (step S109). For shaping, for example, it is possible to convert the inflection form to the end form or the stem, or to the standard form in consideration of the fluctuation of the notation. In the present embodiment, the first search query Q1 is a natural language in text format, but is not limited to this. For example, the first search query Q1 may be in an audio format, or may be a string of words or phrases. The first search query Q1 may be directly input from an input unit (not shown) of the search device 100, or may be input from a separately provided user terminal and sent to the search device 100 via a network such as the Internet. It may be entered.

  FIG. 5 is a diagram illustrating a specific example of processing by the extraction unit 1. When “Nihonbashi's delicious and not so expensive sushi restaurant” is input as the first search query Q1, the extraction unit 1 decomposes the first search query Q1 into morpheme strings by morphological analysis. As a result, the first search query Q1 is “Nihonbashi” “No” “Delicious” “Te” “So much” “High” “No” “O” “Sushi” “Ya” “San”. Next, the stop words “no” “te” “so” “o” “ya” “san” are removed. Then, “high” and “not” are combined into “not high” by chunking, and “delicious” becomes “delicious” by shaping. As a result, the query elements “Nihonbashi”, “delicious”, “not expensive”, and “sushi” are extracted from the first search query Q1.

  FIG. 6 is a diagram illustrating a first example of the processing flow of the calculation unit 2 and the determination unit 3, and corresponds to step S3 and step S5 in FIG. Example 1 is an example in which the determination in the determination unit 3 is performed using a threshold value. The calculation unit 2 acquires one query element qi from the extracted query elements (step S2101), calculates the degree of association f (qi, Dj) between the query element qi and the database Dj, and stores it in, for example, a main memory Store in the device (step S2103). Here, the database Dj is one database among the plurality of databases D1, D2, and D3. A specific method for calculating the degree of association will be described later. The determination unit 3 determines the threshold value of the relevance f (qi, Dj) (step S2105). If the relevance f (qi, Dj) is within the threshold range, it is determined that the query element qi is related to the database Dj, and the query element qi is added to the query list L1dj of the database Dj (step S2107). If the relevance f (qi, Dj) is outside the threshold range in step S2105, the determination unit 3 determines that the query element qi and the database Dj are not related, and the process proceeds to step S2109. In step S2109, it is determined whether the calculation unit 2 has calculated the degree of association with the databases D1, D2, and D3 for the query element qi. In the case of “No”, the calculation unit 2 sets the processing target in the next database (step S2111) and returns to step S2103. By repeating this process, the degree of association with the databases D1, D2, and D3 is calculated for the query element qi, and the presence or absence of the association is determined. If the calculation unit 2 determines “Yes” in step S <b> 2109, the calculation unit 2 determines whether the relevance level has been calculated for all query elements (step S <b> 2113). In the case of “No”, the calculation unit 2 sets the processing target to the next query element (step S2115), and returns to step S2101. By repeating this process, the degree of association with each of the query elements is calculated for each of the query elements, and the presence or absence of the association is determined. The query list L1dj is stored in the query list storage unit 10 in association with the database Dj. The query list L1dj stores a query element whose degree of association with the database Dj is within a threshold range, that is, a column of query elements determined to be associated with the database Dj. Note that it is preferable to set an upper limit as the threshold value in the determination. This is because a query element having an appearance frequency that is too high is likely to be an element used regardless of the type of data, and it is inappropriate to determine that the relation to a specific database is high.

  FIG. 7 is a diagram illustrating a second example of the processing flow of the calculation unit 2 and the determination unit 3, and corresponds to Step S3 and Step S5 of FIG. Example 2 is an example in which the determination in the determination unit 3 is performed based on the highest relevance value. The calculation unit 2 acquires one query element qi from the extracted query elements (step S2201), calculates the degree of association f (qi, Dj) between the query element qi and the database Dj, and stores it in, for example, a main memory Store in the device (step S2203). The calculation unit 2 determines whether the degree of association with the databases D1, D2, and D3 has been calculated for the query element qi (step S2205). In the case of “No”, the calculation unit 2 sets the processing target in the next database (step S2207) and returns to step S2203. By repeating this process, the degree of association f (qi, Dj) with the databases D1, D2, and D3 is calculated for the query element qi. If the determination unit 3 determines “Yes” in step S <b> 2205, the determination unit 3 compares the degrees of association calculated for each of the databases D <b> 1, D <b> 2, and D <b> 3 to identify the database Dj that indicates the highest degree of association. The determination unit 3 determines that the query element qi is related to the database Dj, and adds the query element qi to the query list L1dj of the database Dj (step S2209). The calculation unit 2 determines whether the relevance level has been calculated for all query elements (step S2211). In the case of “No”, the calculation unit 2 sets the processing target to the next query element (step S2213) and returns to step S2201. The query list is stored in the query list storage unit 10. As a result of the processing in Example 2, the query list L1dj of the database Dj includes a query element that has the highest degree of association with the database Dj among the plurality of databases D1, D2, and D3, that is, the association with the database Dj. A column of the determined query elements is stored.

  FIGS. 8A to 8D are diagrams illustrating specific examples of processing performed by the calculation unit 2 and the determination unit 3. When “Nihonbashi”, “delicious”, “not expensive”, and “sushi” are extracted as the query elements (FIG. 8A), the calculation unit 2 sets the query elements “Nihonbashi”, “for each of the databases D1, D2, and D3. The degree of association with “delicious”, “not expensive”, and “sushi” is calculated (FIG. 8B). The determination unit 3 determines whether or not there is a relationship based on the calculated degree of relevance (FIG. 8C). In the process shown in Example 1 above, the determination is performed based on the threshold value. For example, when the threshold is within the range of 10/100 to 50/100, the following occurs. Regarding the database D1, “Nihonbashi” (relevance 10/100) and “sushi” (relevance 20/100) are determined to be related and added to the first query list L1d1 of the database D1 (FIG. 8 ( d)). Regarding the database D2, “delicious” (relevance 30/100) and “not high” (relevance 20/100) are determined to be related and added to the query list L1d2 of the database D2 (FIG. 8D). ). For the database D3, since there is no query element within the threshold range, the query list L1d3 is not created. Note that the relevance values shown in FIG. 8B are set for convenience of explanation and are different from the actual values.

  In the processing shown in Example 2 above, the determination is performed based on the highest degree of association. The determination unit 3 compares the degree of association of the query element “Nihonbashi” with each of the databases D1, D2, and D3, and determines that there is an association with the database D1 having the highest degree of association (FIG. 8C). “Nihonbashi” is added to the first query list L1d1 of the database D1. Similarly, “delicious”, “not expensive”, and “sushi” are also compared, and a query element is added to the first query list L1dj of the database Dj having the highest relevance (FIG. 8D). . As a result, the query elements “Nihonbashi” and “sushi” determined to be related to the database D1 are stored in the query list L1d1, and the query elements “delicious” and “high” determined to be related to the database D2 are stored in the query list L1d2. Is stored.

  FIG. 9 is a diagram illustrating a first example of a processing flow of processing for calculating the relevance, and FIG. 10 is a diagram illustrating the second example. This Example 1 and Example 2 are steps S2103 of the relevance calculation and determination (example 1) processing flow shown in FIG. 6, and the relevance calculation and determination (example 2) processing flow steps shown in FIG. This is processing corresponding to S2203. Example 1 is an example in which the degree of association is calculated based on the appearance frequency. When a specific query element frequently appears in a specific database, the query element and the database are likely to have an association. Therefore, by calculating the degree of association based on the appearance frequency, the degree of association becomes a more appropriate value. The calculation unit 2 acquires one query element qi (step 311), and counts the appearance frequency Rdj of the query element qi in the database Dj (step 313). Next, the calculation unit 2 calculates the degree of association f (qi, Dj) based on the ratio of the appearance frequency Rdj to the number of data elements in the database Dj, and stores it in a storage device such as a main memory (step S315). . The degree of association f (qi, Dj) increases as the appearance frequency increases. Here, the data element of the database Dj is an element corresponding to the query element. For example, the data elements can be extracted by performing similar query element extraction processing by the extraction unit 1 on the data stored in the databases D1, D2, and D3. The number of data elements is the number of data elements extracted from each database. Is obtained by counting the number of. These processes are preferably performed in advance, and the number of data elements is preferably stored in the storage unit for each of the databases D1, D2, and D3. Note that the degree of association does not necessarily need to use a ratio to the number of data elements, and may be calculated using, for example, the appearance frequency Rdj as it is.

  Example 2 shown in FIG. 10 is an example in which the degree of association is calculated using a category. For example, if the location of a store is described in the data, and if the degree of relevance is determined based on a specific place name (such as “Nihonbashi” or “Tokyo”), the specific place name does not appear in the database It is misjudged that the relevance is low. Therefore, in Example 2, the query elements and data elements are replaced with categories, and the relevance is calculated. In the search system 100, a rule for associating a category with a category element included in the category is defined in advance. For example, as shown in FIG. 11, this rule is stored in a storage unit (not shown) of the search system 100 in the form of a category table. When obtaining the query element qi (step S321), the calculation unit 2 refers to the category table and identifies the category ci including the query element qi as a category element (step S323). Next, the calculation unit 2 acquires a category element included in the category ci from the category table, and counts the appearance frequency Rdj in which the category element appears in the database Dj (step S325). Next, the calculation unit 2 calculates the degree of association f (ci, Dj) based on the ratio of the appearance frequency Rdj to the number of data elements in the database Dj, and stores it in a storage device such as a main memory (step S327). Thereby, the degree of association is calculated based on the category, and the degree of association becomes a more appropriate value. Here, the degree of association does not necessarily need to use a ratio to the number of data elements, and may be calculated using, for example, the appearance frequency Rdj as it is.

  As shown in FIG. 11, the category of each data element in the database Dj is preferably specified in advance based on the category table and stored in a storage unit (not shown). Further, the category-related rule may be such that the category is specified by partial matching. For example, in the category table, “bridge”, “town”, “river”, and the like are defined as place name suffixes, and for an element having this place name suffix, the category is determined as “place name”. In addition, as a category, it is a fine-grained category that combines notation fluctuations and synonyms into one category, or a semantic category such as “place name”, “cooking type”, “taste”, “value feeling”, etc. Alternatively, a coarse-grained category of part-of-speech level such as “noun” or “adjective” may be used. The degree of association may be a value that is an index indicating how much the query element and the database are related, and is not limited to Example 1 and Example 2 above. For example, the degree of relevance may be calculated based on the appearance frequency of the character string in units of n character strings (ngram).

  FIG. 12 is a diagram illustrating an example of a processing flow of processing by the search query generation unit 4, and corresponds to Step 7 in FIG. The query generation unit 4 refers to the query list L1dj of the database Dj stored in the query list storage unit 10 (step S701). The query list L1dj stores one or more query elements that are determined to be related to the database Dj. There may be one or more query elements included in the query list L1dj. The query generation unit 4 acquires one or a plurality of query elements stored in the query list L1dj, and generates a second search query Q2dj including the query elements based on a predetermined rule (step S703). Next, the search query generation unit 4 stores the second search query Q2dj in the query storage unit 11 in association with the database Dj (step S705). The search query generation unit 4 determines whether a second search query has been generated for the databases D1, D2, and D3 (step S707). In the case of “No”, the search query generation unit 4 sets the processing target in the query list of the next database (step S709), and returns to step S701. If “Yes”, the process ends. The query storage unit 11 stores second search queries Q2d1, Q2d2, and Q2d3 in association with the databases D1, D2, and D3. As a result, dedicated second search queries Q2d1, Q2d2, and Q2d3 are generated for each of the databases D1, D2, and D3. The second search query Q2dj in the database Dj includes query elements that are determined to be related to the database Dj. However, if there is no query element determined to be related to the database Dj, the second search query Q2dj corresponding to the database Dj is not generated.

  FIGS. 13A and 13B are diagrams illustrating a specific example of processing by the search query generation unit 4. The query generation unit 4 refers to the query list L1d1 of the database D1 stored in the query list storage unit 10, and based on a predetermined rule, includes a second element including the query elements “Nihonbashi” and “sushi” of the query list L1d1. A search query Q2d1 is generated (FIG. 13A). The second search query Q2d1 includes query elements “Nihonbashi” and “sushi” determined to be related to the database D1. Similarly, the search query generation unit 4 refers to the query list L1d2 of the database D2, and generates a second search query Q2d2 including the query elements “delicious” and “not expensive” (FIG. 13B). The second search query Q2d2 includes query elements “delicious” and “not expensive” determined to be related to the database D2. Since there is no query list L1d3 for the database D3, the second search query Q2d3 is not generated. The second search queries Q2d1 and Q2d2 are associated with the databases D1 and D2 and stored in the query storage unit 11. In the example of FIG. 13, the predetermined rule is to connect query elements with “or”, but is not limited thereto.

  FIG. 14A is a diagram illustrating an example of a processing flow of processing by the search units 51, 52, and 53, and corresponds to the processing in step S9 in FIG. The search units 51, 52, and 53 acquire the second query Q2dj stored in the query storage unit 11 (step S5101). Then, when the search unit 51, 52, 53 acquires the second search query Q2dj dedicated to the database Dj in charge, the search unit 51, 52, 53 searches the database Dj in charge based on the second search query Q2dj (step S5103). A data block matching the second search query Q2dj is detected from the database Dj. There may be one detected data block or a plurality of data blocks.

  The first search query Q1 input from the user is a search query common to the databases D1, D2, and D3, and includes query elements that are not related to each database. For this reason, if the databases D1, D2, and D3 are searched based on the first search query Q1, the search accuracy is lowered. On the other hand, the second search query Q2dj is a search query dedicated to the database Dj and includes only query elements that are determined to be related to the database Dj, so that the search accuracy is improved.

  In addition, although the search method in search part 51,52,53 is arbitrary, For example, Boolean search and extended Boolean search are mentioned. A search method may be selected according to the query element. In this case, a table defining the correspondence between query elements and search methods is stored in the storage unit. The search units 51, 52, 53 refer to the table, specify a search method corresponding to the query element included in the second search query Q2dj, and search the database Dj using the search method. Thereby, a search is performed by a search method suitable for the characteristics of the query element, and the search accuracy can be further improved.

  A search method may be selected according to the category of the query element. FIG. 14B is a diagram illustrating a processing flow of processing for selecting a search method according to a category. The search units 51, 52, and 53 acquire a query element from the second search query Q2dj, and based on a category rule that associates a category with a plurality of category elements included in the category, a category including the query element as a category element Is identified (step S5201). Next, the search units 51, 52, and 53 specify the search method corresponding to the specified category based on the search rule that associates the category with the search method (step S5203). Then, the search units 51, 52, and 53 search the database with the search method for the query element. When the second search query Q2dj includes a plurality of query elements, a category is specified for each query element, and a search is performed for each query element using a search method corresponding to the category. Note that the category table described above may be used as the category rule. In addition, the search rule may be stored in a storage unit (not shown) as a table that associates categories with search methods. Thereby, a search is performed by a search method suitable for the characteristics of the category of the query element, and the search accuracy can be further improved.

  Furthermore, the search method may be selected according to the database. In this case, a table in which the databases D1, D2, and D3 are associated with the search method is stored in a storage unit (not shown). The search units 51, 52, and 53 refer to this table, specify a search method corresponding to the responsible database Dj, and search the responsible database Dj by the search method. Thereby, a search is performed by a search method suitable for the characteristics of the database, and the search accuracy can be further improved.

  FIG. 15 is a diagram showing a first example of a processing flow of processing by the set generation units 61, 62, and 63, and FIG. 17 is a diagram showing a second example thereof. Examples 1 and 2 correspond to step S11 in FIG. In Example 1, as a result of the search by the search units 51, 52, and 53, when a data block matching the second search query Q2dj is detected in the database Dj in charge, the set generation units 61, 62, and 63 are detected. The entity Ex associated with the data block is identified (step S1111). There may be one entity or multiple entities. When the entity Ex is specified, the set generation units 61, 62, and 63 add the entity to the entity list L2dj of the database Dj (step S1113). The entity list L2dj includes a column of the entity Ex detected from the database Dj. The set generation units 61, 62, and 63 store the entity list L2dj in the set storage unit 12 in association with the database Dj. The entity list L2dj stored in the set storage unit 12 constitutes a set of entities detected from the database Dj. That is, the set storage unit 12 stores a set of entities detected from each database Dj.

  FIG. 16 is a diagram for explaining a specific example of Example 1 of the processing flow of the set generation units 61, 62, and 63. The inside of the frame A is a description of the processing of the set generation unit 61, the inside of the frame B is the description of the processing of the set generation unit 62, and the inside of the frame C is the description of the processing of the set generation unit 63. Within the frame A, the search unit 51 acquires the search query Q2d1 and searches the database D1. As a result, when the data blocks d1, d2, d3, d4 are detected, the set generation unit 61 identifies the entities E1, E2, E3, E4 associated with the data blocks d1, d2, d3, d4, and the entity list Add to L2d1. Similarly, when the search unit 52 detects the data blocks d5, d6, d7 from the database D2 in the frame B, the set generation unit 62 adds entities E1, E3, E5 associated therewith to the entity list L2d2. Since the dedicated second search query Q2d3 is not generated in the database D3, the processing of the search unit 53 and the set generation unit 63 is not performed. The entity list L2d1 is a set of entities detected from the database D1, and the entity list L2d1 is a set of entities detected from the database D2.

  FIG. 17 is a diagram illustrating a second example of the processing flow by the set generation units 61, 62, and 63. When the search units 51, 52, and 53 detect a data block that matches the second search query Q2dj from the database Dj, the set generation units 61, 62, and 63 select the entity Ex associated with the detected data block. Specify (step S1121). Next, the set generation units 61, 62, and 63 calculate the first evaluation value Vd for each of the detected data blocks (step S1123). The first evaluation value Vd may be a value that represents the degree to which the data block d matches the second search query Q2dj. The higher the matching degree of the data block d with respect to the second search query Q2dj, the higher the first evaluation value Vd. A specific method for calculating the first evaluation value Vd will be described later. The set generation units 61, 62, and 63 associate the entity Ex and the first evaluation value Vd and add them to the entity list L3dj (step S1125). The entity list L3dj is stored in the set storage unit 12 in association with the database Dj.

  FIG. 18 is a diagram illustrating a specific example of Example 2 of the set generation units 61, 62, and 63. Similar to Example 1, the set generation unit 61 identifies entities E1, E2, E3, and E4 associated with the data blocks d1, d2, d3, and d4, and adds them to the entity list L3d1 of the database D1. In addition, the set generation unit 61 calculates a first evaluation value Vd for each of the data blocks d1, d2, d3, and d4 detected by the search unit 51, and associates the entity with the entities E1, E2, E3, and E4. Add to L3d1. The set generation unit 62 performs the same processing, and adds the entity and the first evaluation value Vd to the entity list L3d2 of the database D2. The entity list L3d1 includes a set of columns of entities detected as a result of the search of the database D1 and the first evaluation value Vd. The entity list L3d2 includes a set of columns of entities detected as a result of the search of the database D2 and the first evaluation value Vd. Note that the value of the first evaluation value Vd shown in FIG. 18 is set for convenience of explanation, and is different from the actual value.

  Here, the first evaluation value is calculated as follows. The storage unit (not shown) of the search device 100 stores an evaluation value rule that defines a method for calculating the first evaluation value Vd. The set generation units 61, 62, and 63 refer to the evaluation value rule stored in the storage unit (not shown), and calculate the first evaluation value Vd based on the evaluation value rule. When the second search query Q2dj includes a plurality of query elements, an evaluation value is calculated for each of the plurality of query elements, a sum or an average value of these evaluation values is calculated, and the first block of the data block is calculated. The evaluation value Vd may be used. The following is mentioned as an example of the calculation method of 1st evaluation value Vd. As a first example, there is one that is calculated based on the relationship between the query element of the second search query Q2dj and the data element of the data block. Examples of values representing relevance include similarity, relevance, and approximation. For example, if the query element has a place name attribute, the evaluation value is 1 when the place names are completely matched, and the evaluation value is 0.8 when there is a neighboring town relationship. The calculation method of the value representing the relevance is stored in advance as a calculation rule in a storage unit (not shown) of the search device 100. The set generation units 61, 62, and 63 acquire calculation rules corresponding to the query elements, calculate similarity, relevance, approximation, and the like based on the calculation rules, and store them in a storage device such as a main memory, for example. . For example, in the case of a place name, data indicating the area name and the positional relationship between each area as shown in FIG. 19 is stored in a storage unit (not shown). The set generation units 61, 62, and 63 refer to the data and calculate the first evaluation value Vd based on the positional relationship.

  In addition, the first evaluation value Vd may be calculated using a distance between ontology. For example, when the query element is “sushi”, the set generation units 61, 62, and 63 determine that “tempura”, which is a common Japanese dish, is close, and that French food is far away, and the distance is The corresponding value is set as the first evaluation value Vd. Another method for calculating the first evaluation value is based on the appearance frequency of query elements. A data block in which the appearance frequency of query elements included in the second search query Q2dj is high has a high degree of match with the second search query Q2dj. Therefore, the set generation units 61, 62, and 63 count the appearance frequency of query elements included in the second search query Q2dj for the detected data block, and calculate the first evaluation value Vd based on the appearance frequency. To do. As another calculation method for the first evaluation value Vd, the set generation units 61, 62, and 63 refer to dictionary data such as a thesaurus, and the detected data block has a certain relationship with the query element. The appearance frequency of words (synonyms, antonyms, synonyms, etc.) may be counted, and the first evaluation value Vd may be calculated based on the appearance frequency. The higher the appearance frequency, the higher the first evaluation value Vd.

  FIG. 20 is a diagram illustrating Example 1 of a processing flow of processing by the subset generation unit 7, and FIGS. 22 and 23 are diagrams illustrating Example 2 thereof. Examples 1 and 2 are processes corresponding to step S13 in FIG. Example 1 is a subsequent process when the processing of the set generation units 61, 62, and 63 is set as the processing flow example 1 (FIG. 15). Example 2 is an example of the processing flow of the processing of the set generation units 61, 62, and 63. This is subsequent processing when 2 (FIG. 17).

  In Example 1 shown in FIG. 20, the subset generation unit 7 acquires the entity list L2dj of the database Dj from the set storage unit 12 (step 13101). Next, the subset generation unit 7 acquires the entity Ex from the entity list L2dj (step S13103), and adds the database Dj to the evaluation list L4ex of the entity Ex (step S13105). There may be one entity or multiple entities. The subset generation unit 7 determines whether all entities E included in the entity list L2dj have been processed (step S13107). In the case of “No”, the subset generation unit 7 sets the processing target to the next entity (step S13109) and returns to step S13103. By repeating this process, the database Dj is added to the corresponding evaluation list L4ex for all the entities Ex included in the entity list L2dj. If “Yes” is determined in step S13107, the subset generation unit 7 determines whether all entity lists have been processed (step S13111). In the case of “No”, the subset generation unit 7 sets the processing target to the next entity list (step S13113) and returns to step S13101. Thereby, an evaluation list is generated for each of the entities. The generated evaluation list is stored in a storage device such as a main memory, for example. The evaluation list L4ex of the entity Ex is provided with a database column in which the entity Ex is detected. The subset generation unit 7 refers to the generated evaluation list of each entity, extracts an entity based on a predetermined rule, and stores the extracted entity list L5 in the subset storage unit 13 (step S13115). . The list L5 stored in the subset storage unit 13 constitutes a subset of entities.

  Examples of the predetermined rule when extracting an entity include the following. In the search by the search units 51, 52, and 53, there is a high possibility that an entity detected in many databases is an entity that matches the first search query Q1. Therefore, for example, the subset generation unit 7 may extract the entity Ex when the number of databases included in the evaluation list L4ex of the entity Ex is equal to or greater than a threshold value. In another example, the subset generation unit 7 regards the entity list generated by the set generation units 61, 62, and 63 as one set, and generates the union or intersection set to generate the subset. It may be generated.

  FIGS. 21A to 21C are diagrams illustrating a specific example of the processing flow example 1 of processing by the subset generation unit 7. As illustrated in FIG. 21A, the subset generation unit 7 acquires the entity E1 from the entity list L2d1, and adds the database D1 to the evaluation list L4e1 of the entity E1. Similarly, the subset generation unit 7 acquires entities E2, E3, E4 from the entity list L2d1, and adds the database D1 to the evaluation lists L4e2, L4e3, L4e4. Further, as illustrated in FIG. 21B, the subset generation unit 7 acquires the entities E1, E3, and E5 from the entity list L2d2 and adds them to the entity evaluation lists L4e1, L4e3, and L4e5. The evaluation lists L4e1 to L4e5 are stored in a storage device such as a main memory, for example. The subset generation unit 7 extracts entities from the entities E1, E2, E3, E4, and E5 based on a predetermined rule. In this specific example, when two or more databases are included in the evaluation list L4ex of the entity Ex, the entity Ex is extracted. As shown in FIG. 21C, the subset generation unit 7 extracts entities E1 and E3 whose evaluation list includes two or more databases and adds them to the list L5. List L5 is a subset of entities.

  Example 2 shown in FIGS. 22 and 23 is a subsequent process when the processing of the set generation units 61, 62, and 63 is set as Example 2 (FIG. 17) of the processing flow. The following points are different from Example 1 shown in FIG. Note that a description of points in common with Example 1 is omitted. In step S13203, the subset generation unit 7 acquires the entity Ex and the first evaluation value Vd from the entity list L3dj. In step 13205, the subset generation unit 7 adds the first evaluation value Vd to the evaluation list L4ex of the entity Ex. The evaluation list L4ex is associated with an entity and stored in a storage device such as a main memory, for example. In Example 1, the database Dj is added to the evaluation list L4ex. However, in this example, the first evaluation value Vd is added. As illustrated in FIG. 23, the subset generation unit 7 calculates the comprehensive evaluation value Vx of the entity Ex based on the first evaluation value Vd included in the evaluation list L4ex of the entity Ex (step S13215). The comprehensive evaluation value Vx is associated with the entity Ex and stored in a storage device such as a main memory, for example. The comprehensive evaluation value Vx represents the degree of match of the entity Ex with respect to the first query Q1, and the higher the degree of match, the higher the comprehensive evaluation value Vx. When a plurality of first evaluation values Vd are stored in the evaluation value list L4ex of the entity Ex, the total or average value of the first evaluation values Vd may be used as the comprehensive evaluation value. The subset generation unit 7 determines whether or not the comprehensive evaluation values of all entities have been calculated (step S13217). In the case of “No”, the subset generation unit 7 sets the processing target to the next entity (step S13219) and returns to step S13215. By repeating this process, comprehensive evaluation values are calculated for all entities. If “Yes” is determined in step S13217, the subset generation unit 7 refers to the comprehensive evaluation value of each entity and extracts the entity based on a predetermined rule. The extracted entity list L5 is stored in the subset storage unit 13 (step S13221). Here, as the predetermined rule, for example, a rule for extracting an entity whose comprehensive evaluation value is equal to or greater than a threshold value can be cited. The list L5 stored in the subset storage unit 13 is a subset of entities.

  24A to 24C are diagrams illustrating a specific example of the second example. As illustrated in FIG. 24A, the subset generation unit 7 acquires the first evaluation value Vd of each entity E1 to E4 from the entity list L3d1, and adds it to the evaluation list L4e1 to L4e4 of each entity. Further, as illustrated in FIG. 24B, the subset generation unit 7 acquires the first evaluation value Vd of each entity E1, E3, E5 from the entity list L3d2, and the evaluation list L4e1, L4e3 of each entity. , L4e5. The evaluation list is associated with the entity and stored in a storage device such as a main memory. The subset generation unit 7 calculates the total sum of the first evaluation values Vd for each of the evaluation lists L4ex, and sets the total evaluation values V1 to V5 of the entities E1 to E5. The comprehensive evaluation value is associated with the entity and stored in a storage device such as a main memory, for example. As illustrated in FIG. 24C, the subset generation unit 7 extracts entities based on a predetermined rule and adds them to the list L5. The list L5 is stored in the subset storage unit 13. Here, the entity E1 and the entity E3 whose comprehensive evaluation values are equal to or greater than the threshold are extracted. List L5 is a subset of entities.

  The generated subset of entities is output to an output device such as a monitor or a printer and provided to the user. Thereby, the user can obtain a subset of entities as a search result of the first search query Q1. The subset of entities may be provided as a list of entities or as a list of links to data blocks associated with the entities. In the present embodiment, search results are not output for each database, but search results for each database are integrated and output. The output entity is determined to have a high degree of match with the first search query Q1 based on the comprehensive evaluation value. Therefore, the user can obtain an entity close to information needs.

  Moreover, the search apparatus 200 is mentioned as another example of embodiment of this technique. FIG. 25 is a functional block diagram illustrating the search device 200. In FIG. 25, the same elements as those of the search device 100 are denoted by the same reference numerals, and the description thereof is omitted. The search device 200 does not include the set generation units 61, 62, and 63, the set storage unit 12, the subset generation unit 7, and the subset storage unit 13 in the search device 100. That is, the search device 200 performs processing up to the search units 51, 52, and 53 of the search device 100. As a search result, a list of detected data blocks and a list of links to the detected data blocks are output. The search device 100 searches for an entity associated with a data block, but the search device 200 is effective when it is desired to search the data block itself.

  Moreover, the search apparatus 300 is mentioned as another example of embodiment of this technique. FIG. 26 is a functional block diagram illustrating the search device 300. In FIG. 26, the same elements as those of the search device 100 are denoted by the same reference numerals, and the description thereof is omitted. The search device 300 does not include the subset generation unit 7 and the subset storage unit 13 in the search device 100. That is, the search device 300 performs processing up to the set generation units 61, 62, and 63. As a search result, an entity list stored in the set storage unit 12 is output, or a link to an entity included in the entity list is output. Since the entity list is generated for each database, the search result is output for each database. The search device 100 integrates the search results of each database and outputs a subset of entities. The search device 300 is effective when it is desired to obtain search results separately for each database.

  Although the embodiment of the present technology has been described above, the present technology is not limited to this. For example, the functional block diagrams of FIGS. 2, 25, and 26 are examples, and do not necessarily match the actual program module configuration. As for the processing flow, as long as the processing result does not change, the order of the steps may be changed or may be executed in parallel.

  The search devices 100, 200, and 300 described above are computer devices, and as shown in FIG. 27, a memory 2501, a CPU (Central Processing Unit) 2503, and a hard disk drive (HDD: Hard Disk Drive) 2505. A display control unit 2507 connected to the display device 2509, a drive device 2513 for the removable disk 2511, an input device 2515, and a communication control unit 2517 for connecting to a network are connected by a bus 2519. An operating system (OS) and an application program for executing the processing in this embodiment are stored in the HDD 2505, and are read from the HDD 2505 to the memory 2501 when executed by the CPU 2503. The CPU 2503 controls the display control unit 2507, the communication control unit 2517, and the drive device 2513 according to the processing content of the application program, and performs a predetermined operation. Further, data in the middle of processing is mainly stored in the memory 2501, but may be stored in the HDD 2505. In an embodiment of the present technology, an application program for performing the above-described processing is stored in a computer-readable removable disk 2511 and distributed, and installed from the drive device 2513 to the HDD 2505. In some cases, the HDD 2505 may be installed via a network such as the Internet and the communication control unit 2517. Such a computer apparatus realizes various functions as described above by organically cooperating hardware such as the CPU 2503 and the memory 2501 described above and programs such as the OS and application programs. .

  The above-described embodiment can be summarized as follows.

  In the search method according to the present embodiment, (A) one or more query elements included in the first search query input by the user are extracted, and (B) different types of data are obtained for each query element. Calculating the degree of association with a plurality of databases to be stored; (C) determining whether or not there is an association with each of the plurality of databases for each query element based on the degree of association; and (D) for each of the plurality of databases. If there is a query element determined to be related to the database, a second search query including the query element is generated. (E) Based on the second search query, among the plurality of databases, A process of searching a database corresponding to the second search query. Here, the plurality of databases may be physically separated databases as shown in FIG. 1A and FIG. 1B, or as shown in FIG. The databases may be physically the same and virtually separated.

  According to this search method, when a user inputs a first search query as a search query common to a plurality of databases, a second search query dedicated to the database is generated for each database. The second search query includes a query element that is determined to be related to the database in charge. Therefore, the user does not have to input a search query for each database or think about a search query suitable for the type of data in the database, which increases convenience. The first search query is common to a plurality of databases, but this does not reduce the search accuracy. The user can obtain highly accurate search results without inputting a dedicated search query to each database. Regardless of what query elements are included in the search query, the degree of relevance is determined for each query element, and whether or not there is an association with each database is determined, so the degree of freedom of the first search query is also high.

  In addition, a plurality of data blocks associated with one entity may be distributed and stored in the plurality of databases. In this case, in the search process described above, (E1) when a data block is detected from each of two or more databases among a plurality of databases, the entity associated with the data block is specified, There is a case where a set of entities is generated, and (E2) a process of generating a subset of the entities from the set of entities based on a predetermined rule may be executed. Accordingly, even if a data block is detected from each of two or more databases, an integrated search result can be obtained by using a subset of entities based on a predetermined rule. Since the search result is a list of entities, the user can obtain a search result suitable for information needs. Here, the predetermined rule may be any one that integrates a set of entities obtained from two or more databases into one set. For example, it may be a union or intersection of sets of entities obtained from each database. Further, for each entity, a comprehensive evaluation value indicating the degree of conformity to the first search query may be calculated, and only entities having a total evaluation value equal to or greater than a threshold may be extracted and set as a subset. Here, the data block may be physically separated as shown in FIG. 1 (a) or 1 (b), or may be physically separated as shown in FIG. 1 (c). They may be the same and virtually separated.

  Further, in the process of calculating the relevance described above, for each query element, the frequency of appearance of the query element in each of the plurality of databases is counted, and based on the frequency of appearance, the query element and each of the databases are counted. It is also possible to calculate the degree of relevance. When a specific query element frequently appears in a specific database, the query element and the database are likely to have an association. In this way, the degree of association becomes a more appropriate value.

  Furthermore, the process of calculating the relevance described above specifies (B1) a category including a query element as a category element based on a category rule that associates a category with a plurality of category elements included in the category, B2) Acquire a plurality of category elements included in the specified category based on the category rule, (B3) Count the appearance frequency of the acquired plurality of category elements for each of the plurality of databases, and (B4) Appearance A process of calculating the degree of association between the query element and each of the plurality of databases based on the frequency may be included. For example, if the location of a store is described in the data, and if the degree of relevance is determined based on a specific place name (such as “Nihonbashi” or “Tokyo”), the specific place name does not appear in the database It is erroneously determined that the relevance is low. In this way, the degree of association is calculated based on the category, and the degree of association becomes a more appropriate value.

  Further, the search processing described above specifies (E3) a category including a query element as a category element based on a category rule that associates a category with a plurality of category elements included in the category, and (E4) a category And (E5) a process of generating a second search query including a query element based on the specified search method, specifying a search method corresponding to the specified category based on a search rule that associates the search method with It may be included. In this way, the search is performed by the search method corresponding to the category, and the search accuracy is further increased.

  Further, the process for generating the set of entities described above performs a first evaluation indicating a degree of matching with the second search query based on a predetermined rule for each data block detected in the search process. You may make it include the process which calculates a value. At that time, the process of generating a subset of the entities described above calculates, for each entity, a comprehensive evaluation value indicating the evaluation of the entity based on the first evaluation value of the data block associated with the entity. In addition, a process of generating a subset of entities from the set of entities based on the comprehensive evaluation value may be included. In this way, the first evaluation value of the data block having a high degree of match with the first search query is increased. And based on the 1st evaluation value of a data block, the comprehensive evaluation value of the entity linked | related with the data block is calculated. The overall evaluation value increases as the degree of matching between the entity and the first search query increases. The subset includes an entity having a high degree of matching with the first search query, and the user can obtain a search result that matches the information needs.

  The program is stored in a computer-readable storage medium or storage device such as a flexible disk, an optical disk such as a CD-ROM, a magneto-optical disk, a semiconductor memory (for example, ROM), or a hard disk. Note that data being processed is temporarily stored in a storage device such as a RAM.

  The following supplementary notes are further disclosed with respect to the embodiments including the above examples.

(Appendix 1)
Extracting one or more query elements included in a first search query input by a user;
For each of the query elements, calculate the degree of association with a plurality of databases each storing different types of data,
Based on the degree of association, for each query element, determine whether or not there is an association with each of the plurality of databases,
For each of the plurality of databases, when there is a query element determined to be related to the database, a second search query including the query element is generated, and the plurality of the plurality of databases are based on the second search query. A search program for causing a computer to execute a process of searching a database corresponding to the second search query among the databases.

(Appendix 2)
A plurality of data blocks associated with one entity are distributed and stored in the plurality of databases;
In the search process, when the data block is detected from each of two or more databases among the plurality of databases, an entity associated with the data block is specified, and a set of the entities is generated.
The search program according to appendix 1, for causing the computer to further execute a process of generating a subset of the entity from the entity set based on a predetermined rule.

(Appendix 3)
The process of calculating the relevance is
For each query element, count the frequency of occurrence of the query element in each of the plurality of databases,
The search program according to appendix 1 or 2, including a process of calculating the degree of association between the query element and each of the databases based on the appearance frequency.

(Appendix 4)
The process of calculating the relevance is
Based on a category rule that associates a category with a plurality of category elements included in the category, a category including the query element as the category element is identified,
Obtaining the plurality of category elements included in the identified category based on the category rule;
For each of the plurality of databases, count the frequency of appearance of the acquired category elements,
The search program according to appendix 1 or 2, including a process of calculating the degree of association between the query element and each of the plurality of databases based on the appearance frequency.

(Appendix 5)
The process of searching includes
Based on a category rule that associates a category with a plurality of category elements included in the category, a category including the query element as a category element is identified,
Based on a search rule that associates the category with a search method, a search method corresponding to the specified category is specified,
The search program according to appendix 1 or 2, including a process of searching based on the specified search method.

(Appendix 6)
The process of generating the set of entities includes
For each data block detected in the search process, calculate a first evaluation value indicating a degree of match with the second search query based on a predetermined rule;
The process of generating a subset of the entities comprises
For each of the entities, based on the first evaluation value of the data block associated with the entity, a comprehensive evaluation value indicating the evaluation of the entity is calculated.
The search program according to claim 2, further comprising: generating a subset of the entities from the set of entities based on the comprehensive evaluation value.

(Appendix 7)
Extracting one or more query elements included in the first search query input by the user;
For each of the query elements, calculate the degree of association with a plurality of databases each storing different types of data,
Based on the degree of association, for each query element to determine the presence or absence of association with the plurality of databases,
For each of the plurality of databases, if there is a query element determined to be related to the database, a second search query including the query element is generated,
A search method executed by a computer, including a process of searching a database corresponding to the second search query among the plurality of databases based on the second search query.

(Appendix 8)
An extraction unit that extracts one or more query elements included in the first search query input by the user;
A calculation unit that calculates a degree of association with each of a plurality of databases that store different types of data for each of the query elements;
A determination unit that determines presence / absence of association with each of the plurality of databases for each query element based on the degree of association;
For each of the plurality of databases, when there is the query element determined to be associated with the database, a search query generation unit that generates a second search query including the query element;
A search device comprising: a search unit that searches a database corresponding to the second search query among the plurality of databases based on the second search query.

100, 200, 300 Search device 1 Extraction unit 2 Calculation unit 3 Determination unit 4 Search query generation units 51, 52, 53 Search units 61, 62, 63 Set generation unit 7 Subset generation unit 10 Query list storage unit 11 Query storage unit 12 Set storage unit 13 Subset storage unit D1, D2, D3 Database

Claims (8)

Extracting one or more query elements included in a first search query input by a user;
For each of the query elements, calculate the degree of association with a plurality of databases each storing different types of data,
Based on the degree of association, for each query element, determine whether or not there is an association with each of the plurality of databases,
For each of the plurality of databases, if there is a query element determined to be related to the database, a second search query including the query element is generated,
A search program for causing a computer to execute a process of searching a database corresponding to the second search query among the plurality of databases based on the second search query. A plurality of data blocks associated with one entity are distributed and stored in the plurality of databases;
In the search process, when the data block is detected from each of two or more databases among the plurality of databases, an entity associated with the data block is specified, and a set of the entities is generated.
The search program according to claim 1, further causing the computer to execute a process of generating a subset of the entity from the entity set based on a predetermined rule. The process of calculating the relevance is
For each of the query elements, a process of counting the appearance frequency of the query element in each of the plurality of databases and calculating the degree of association between the query element and each of the databases based on the appearance frequency. The search program according to claim 1 or 2. The process of calculating the relevance is
Based on a category rule that associates a category with a plurality of category elements included in the category, a category including the query element as the category element is identified,
Obtaining the plurality of category elements included in the identified category based on the category rule;
For each of the plurality of databases, count the frequency of appearance of the acquired category elements,
The search program according to claim 1, further comprising: calculating the degree of association between the query element and each of the plurality of databases based on the appearance frequency. The process of searching includes
Based on a category rule that associates a category with a plurality of category elements included in the category, a category including the query element as a category element is identified,
Based on a search rule that associates the category with a search method, a search method corresponding to the specified category is specified,
The search program according to claim 1, further comprising a process of searching based on the specified search method. The process of generating the set of entities includes
For each data block detected in the search process, calculate a first evaluation value indicating a degree of match with the second search query based on a predetermined rule;
The process of generating a subset of the entities calculates, for each entity, a comprehensive evaluation value indicating an evaluation of the entity based on the first evaluation value of the data block associated with the entity;
The search program according to claim 2, further comprising: generating a subset of the entities from the set of entities based on the comprehensive evaluation value. Extracting one or more query elements included in the first search query input by the user;
For each of the query elements, calculate the degree of association with a plurality of databases each storing different types of data,
Based on the degree of association, for each query element to determine the presence or absence of association with the plurality of databases,
For each of the plurality of databases, if there is a query element determined to be related to the database, a second search query including the query element is generated,
A search method executed by a computer, including a process of searching a database corresponding to the second search query among the plurality of databases based on the second search query. An extraction unit that extracts one or more query elements included in the first search query input by the user;
A calculation unit that calculates the degree of association with each of a plurality of databases that store different types of data for each of the query elements;
A determination unit that determines presence / absence of association with each of the plurality of databases for each query element based on the degree of association;
For each of the plurality of databases, when there is the query element determined to be associated with the database, a search query generation unit that generates a second search query including the query element;
A search device comprising: a search unit that searches a database corresponding to the second search query among the plurality of databases based on the second search query.

Images