JP4942350B2 - Search client - Google Patents

Description

  The present invention relates to a search technique, and more particularly to a search client that uses a server client type index.

  There is a search technique as a technique for obtaining desired information from a large amount of documents. This search technique compares a character string or word shown to a user with a character string or word in a document, and presents a completely matched document or a document with a high degree of matching to the user.

  In search technology, the presence or absence of an index affects the search processing speed. In order to search for specific information, the index is an arrangement of words or symbols indicating the information in a certain sequential order to indicate the location of the information.

  For example, when searching for a document including the word A, if the index is created in advance, the word A may be checked in the index. However, if the index is not created, it is necessary to check the existence of the word A for all documents. Yes, it takes time.

  For this reason, an index is generally held in a server client type search system that performs a plurality of search processes simultaneously at high speed.

  On the other hand, not only in Japanese but also in natural language, there are notation fluctuations and synonyms. As for the shaking of the notation, it appears in character units such as uppercase / lowercase letters, half-width / full-width, English singular / plural types, etc., which appears grammatically in units of words, katakana, foreign words Some words appear irregularly on a word basis. As for synonyms, there are things like "electronic computer" and "computer" that have the same meaning but different notations. These wobbles and synonyms are the cause of search omission and adversely affect the search.

  For example, in a search system having an index, when an index search is performed with a search word to be searched as “computer”, the initial letter “Computer” does not hit. If the search term is “computer”, “computer” will not be hit.

  For this reason, in many cases, normalization processing is performed at the time of creating an index to reduce such a search omission.

  A document group having a normalized index as described above can be searched at a high speed with few leaks. On the other hand, if a change occurs in the normalization processing method, the index must be recreated. There is a problem of not becoming. In addition, an index that has been created once after normalization cannot reflect user-specific characteristics and is not convenient.

  As a method for solving such problems, there is a method in which normalization processing is not performed and all search terms are expanded into possible variations at the time of search. In this method, the expanded word group is enormous. This is not appropriate when processing speed is required.

  For example, the following documents can be cited as conventional documents related to search technology.

  Japanese Patent Laid-Open No. 2001-236358 (Patent Document 1) includes a method that can specify a normalization level for a search word and an index word and a method that can specify expansion of the search word to related notations according to the user's needs. Thus, a document search method and apparatus capable of realizing a trade-off between search accuracy, search efficiency, reproduction rate, and relevance rate are disclosed.

Japanese Patent Laid-Open No. 2002-230021 (Patent Document 2) discloses information with high search accuracy according to a user's intention by analyzing a search query according to co-occurrence information and solving ambiguity in a similar document search using a vector space model. An information retrieval device, an information retrieval method, and a storage medium capable of performing retrieval are disclosed.
JP 2001-236358 A Japanese Patent Laid-Open No. 2002-230021

  However, the invention disclosed in Patent Document 1 cannot reflect user-specific characteristics, and the invention disclosed in Patent Document 2 has a problem that the search processing speed cannot be increased.

  The present invention has been made in order to solve these problems, and a search client capable of executing high-precision search processing at high speed and reflecting user-specific characteristics is provided. It is intended to provide.

  In order to achieve the above object, the image forming apparatus of the present invention employs the following configuration.

Search client, and the standard index generated by normalized string using basic words data and a standard dictionary that the regularization result when parsing language, search characters included in the search request A search for accessing a search server having a character unit normalization index generated by a character string normalized using a character normalization dictionary for performing character unit normalization processing on the column, and performing a search In the client, the search client is capable of additionally registering user-specific words, an editable user dictionary collecting special word data for each individual, the standard dictionary, the character normalization dictionary, Search command generation means for generating a search command, wherein the search command generation means stores the search character string in the standard dictionary and the user dictionary, respectively. Independent search, a standard search result set that is a set of character strings whose representative words are the character strings of search results in the standard dictionary, and a set of character strings whose representative words are the character strings of search results in the user dictionary A user search result set is extracted, a logical operation process is performed on the elements of the standard search result set and the elements of the user search result set to generate a search command, and the search command is sent to the search server It can be set as the structure which transmits.

  As a result, it is possible to provide a search client that can suppress a decrease in search processing speed, reflect user-specific characteristics, reduce search omissions, and execute search processing with higher accuracy.

Also, search client, further the search command generating means, said a standard search results in the set the user search results are not included in the set first set of character strings is included in the user search result set the A configuration in which a second character string set not included in the standard search result set is extracted, and a logical operation process is performed on the first character string set and the second character string set to generate the search command It can be.

  Accordingly, it is possible to provide a search client that can execute a search process that reflects user-specific characteristics.

In order to achieve the above object, the search client of the present invention uses the standard word data used for language analysis and the character strings normalized using n standard dictionaries that collect the normalization results. Normalization is performed using the character normalization dictionary for performing character-by-character normalization processing on the n standard indexes corresponding to the n standard dictionaries and the search character string included in the search request. In a search client that performs a search by accessing a search server having a character-unit normalized index generated by a converted character string, the search client can additionally register user-specific words, and editable user dictionary that special word data for each individual, the n number of the standard dictionary, and the character normalization dictionary, a search command generating means for generating a search command, the Includes a number of operations detecting means for detecting the number of operations of the logical operations involved in the search command, and the search command generating means, the user as one of the standard dictionary of the search string the n standard dictionary searches independently in the dictionary, and the set der Ru collection if Xi string typified word string search result of the one of the standard dictionary, typified word string of the search results in the user dictionary extracting a set der Ru collection case Yi strings, the first set of results obtained by searching the standard index that corresponds with the one of the standard dictionary elements of the set Xi search string, the set A set of results obtained by searching the character unit normalized index with each element constituting the set SUBi included in Xi and not included in the set Yi, and a set ADDi included in the set Yi and not included in the set Xi Constitute Search comprising logical operation to extract a second set of results the logical sum of the set of results obtained by searching the character unit normalization index in element, the first set that are not included in logical product of The search client generates a command of the logical operation detected by the operation count detection unit among the search commands generated corresponding to each of the n standard dictionaries in the search command generation unit. A search command with the smallest number of operations can be transmitted to the search server.

  As a result, it is possible to provide a search client that can suppress a decrease in search processing speed, reflect user-specific characteristics, reduce search omissions, and execute search processing with higher accuracy.

Further, it is possible to provide a search client capable of reducing the load on the search server by generating a search command with a small number of calculations.

  In order to achieve the above object, the search client of the present invention further includes a standard dictionary update unit for updating the standard dictionary based on the standard dictionary update data distributed from the search server. Can do.

  Thereby, even when the standard dictionary is changed, the standard dictionary can be updated based on the standard dictionary update data. Thereby, the effect obtained by the present invention can be continued. In other words, it is possible to provide a search client that can suppress a decrease in search processing speed, reflect user-specific characteristics, reduce search omissions, and execute search processing with higher accuracy.

  According to the search client of the present invention, it is possible to execute search processing with higher accuracy and reflect user-specific characteristics while suppressing a decrease in search processing speed.

  Embodiments of the present invention will be described below with reference to the drawings.

  When performing a search process, the search client of the present invention sends a search command generated based on a user dictionary editable by a user and a standard dictionary to the search server, and causes the search server to execute the search process. .

  FIG. 1 is an example of a functional block diagram of a search system having a search client according to the first embodiment of the present invention.

  The search system 10 includes a search server 20 and a search client 30, and the search server 20 and the search client 30 are connected via a network.

  The search system 10 performs a search process. In the search client 30, a search command is generated based on a search request input from a user, and the search command is sent from the search client 30 to the search server 20. The search server 20 receives this and performs a search process.

  Here, the search request is input from the user and represents a search target that the user performs a search process in the search system 10 as a character string. This character string becomes a search character string to be searched in the present embodiment.

  The search server 20 includes an index A1, an index B, a document database 26 (hereinafter, document DB 26), and a standard dictionary distribution unit 28. The search server 20 receives the search command sent from the search client 30 and performs a process for searching the index A1 and the index B. As a result, the position information (location information) of the document desired by the user existing in the document DB 26 is obtained. ) The position information is sent to the search client 30 as a search result. The user can easily browse a desired document stored in the document DB 26 based on the position information of the document as the search result.

  The index A1 is an index generated by a character string that has been subjected to a standard normalization process using a standard dictionary. Here, the standard dictionary is, for example, a collection of basic word data and its normalization results when a language is analyzed. In addition to the normalization result of the document data stored in the document DB 26 using the standard dictionary, the index A1 includes attributes such as the name and size of the document data, and URL that is position information indicating the recording location of the document data. (Uniform Resource Locator) and the like are held.

  The index B is an index generated by a character string that has been subjected to normalization processing using a character normalization dictionary that performs only character-unit normalization. In addition, the index B includes attributes such as the name and size of the document data, and position information indicating the recording location of the document data, in addition to the result of the character normalization dictionary of the document data stored in the document DB 26. A certain URL or the like is held.

  In the above, since information other than the normalization result is the same among the information held in the index A1 and the index B, information other than the normalization result is shared and held in the index A1 and the index B. A specific method of sharing is shown in a CREATE statement by SQL described later.

  A large number of document data is stored in the document DB 26 and can be traced from each of the index A1 and the index B. In this embodiment, the document DB 26 is stored in the search server 20, but the document DB 26 may be connected to the outside of the search server 20 by an appropriate connection method.

  The standard dictionary distribution unit 28 distributes a new standard dictionary to the search client 30 through the network when the normalization processing method of the index A1 is changed due to a change of the standard dictionary or the like.

  The search client 30 includes a search command generation unit 32, a standard dictionary update unit 34, a user dictionary 36, a standard dictionary 38, and a character normalization dictionary 39. When a search character string is input as a search request from the user, the search client 30 generates a search command and sends the search command to the search server 20.

  The search command generation unit 32 expands the search character string into related terms using the user dictionary 36 and the standard dictionary 38, and generates a search command based on the result.

  The standard dictionary update unit 34 updates the standard dictionary 38 based on the new standard dictionary distributed from the standard dictionary distribution unit 28.

  The user dictionary 36 is a collection of special word data for each individual, allowing the user to additionally register words unique to the user.

  The standard dictionary 38 is a dictionary used for generating the index A1 in the search server 20. The standard dictionary 38 is a collection of basic word data when analyzing languages, for example, and is used for comparison with a user dictionary.

  The character normalization dictionary 39 is a dictionary used for generating the index B in the search server 20. The character normalization dictionary 39 is used to perform a character unit normalization process on the search character string included in the search request.

  Here, in the present embodiment, the search server 20 receives the SQL command. In the search server 20, a table showing the attribute of the document data stored in the document DB 26 and the URL that is the position information of the document data is created in advance. A command indicating creation of a table is “create table table1 (document URI text, IA1 text, IB text)”. Here, “IA1” indicates the index A1, and “IB” indicates the index B.

  Further, the index A1 and the index B in the search server 20 are generated as follows.

  In the search server 20, a character string registered in the index A1 and the index B is extracted as text T1 from the document input by the user. The extracted text T1 is first subjected to normalization processing by a character normalization dictionary, and is registered in the index B as a character string subjected to character-unit normalization processing. Here, the text T1 subjected to normalization processing by the character normalization dictionary is indicated as “IB1”.

  At this time, in the index B, IB1 is registered in association with the position information URL1 indicating the recording location of the character string input by the user.

  This IB1 is further subjected to normalization processing by a standard dictionary, and is registered in the index A1 as a character string subjected to standard normalization processing. Here, the text IB1 subjected to normalization processing by the standard dictionary is indicated as “IA11”. At this time, in the index A1, like the index B, the IA11 and the positional information URL1 indicating the recording location of the character string input by the user are registered in association with each other.

  As a result, the index A1 and the index B store character strings subjected to the above-described normalization processes.

  FIG. 2 is a flowchart showing the flow of search processing in the search system 10.

  When the user inputs a search request to the search client 30, the search command generation unit 32 receives the search request and generates a search command (S210). Here, the search command generation method changes according to the input search request. In this embodiment, a search request is formed by concatenating a plurality of search character strings with spaces.

  That is, the search request in S210 is obtained by concatenating character string a, character string b, character string c, and character string d with a blank. The character string a, the character string b, the character string c, and the character string d correspond to search character strings.

  The search command generation unit 32 generates a search command “select * from table1 where A and B and C and D” from the search request “a b c d”. Here, A, B, C, and D are partial search commands for performing search processing for each search character string in the search server 20. The partial search command is generated for each search character string. Therefore, the search command generation unit 32 generates the same number of partial search commands as the number of search character strings included in the search request. The partial search command A, partial search command B, partial search command C, and partial search command D will be specifically described below.

  The partial search command A, the partial search command B, the partial search command C, and the partial search command D are searched using the index W with the character string a, the character string b, the character string c, and the character string d as search character strings. Is shown. Here, although there is no index W, it is replaced with the index A1 or index B by a user dictionary expansion process described later, and is used for the explanation of the partial search command for convenience.

  That is, the partial search command A is a partial search command using the character string a as a search character string, and is indicated as “IW like“% nB (a)% ””. The partial search command B is a partial search command using the character string b as a search character string, and is indicated as “IW like‘% nB (b)% ’”. The partial search command C is a partial search command using the character string c as a search character string, and is indicated as “IW like '% nB (c)%'”. The partial search command D is a partial search command using the character string d as a search character string, and is indicated as “IW like '% nB (d)%'”.

  Here, IW in each partial search command indicates the index W, and nB (a), nB (b), nB (c), and nB (d) are character strings a that are search character strings. The character string after normalizing the character string b, the character string c, and the character string d using the character normalization dictionary 39 is shown.

  Next, the search command generation unit 32 performs a user dictionary expansion process on each search character string included in the partial search command A, the partial search command B, the partial search command C, and the partial search command D generated in S210. , A search command for transmission “select * from table1 where A ′ and B ′ and C ′ and D ′” is generated (S220).

  Here, A ′, B ′, C ′, and D ′ indicate that the user dictionary expansion processing has been performed for each of the partial search command A, the partial search command B, the partial search command C, and the partial search command D. .

  Details of the processing in S220 will be described later.

  The search command for transmission generated in this way is transmitted to the search server 20 via the network (S230). Then, the sent search command is received by the search server 20, and the search server 20 that has received the search command starts search processing in the index A1 and the index B (S240). By this search processing, the search server 20 obtains position information of the search character string in the document DB 26.

  After the search process is completed, the search server 20 sends this position information as a search result to the search client 30 via the network, and the search client 30 receives this search result (S250). Then, in the search client 30, this search result is displayed to the user. The user can easily browse a desired document stored in the document DB 26 based on the position information of the search character string as the search result.

  Here, the details of the process in S220 will be described with reference to FIG.

  FIG. 3 is a flowchart for explaining the user dictionary expansion processing in S220. In FIG. 3, the search character string included in the partial search command is Si (i = 0, 1, 2,..., M).

  The search command generation unit 32 sets the search character string included in the first partial search command among a plurality of partial search commands as S0, and starts processing (S310). In this embodiment, the first partial search command corresponds to the partial search command A.

  Next, the search command generation unit 32 determines whether the user dictionary expansion process has been performed up to the search character string included in the last partial search command, and is included in the partial search command that has not been subjected to the user dictionary expansion process. The user dictionary expansion process is started for the search character string to be searched (S320). In the present embodiment, the last partial search command corresponds to the partial search command D.

  The search command generator 32 first performs a process of searching the standard dictionary 38 using the search character string Si (S330). As a search result, a character string obtained from the standard dictionary 38 is indicated as a character string nA (Si). Next, the search command generation unit 32 extracts the character string set Xi, which is a set of character strings related to the character string nA (Si), from the standard dictionary 38 with the character string nA (Si) as a representative word. Perform expansion processing. (S340).

  Next, the search command generation unit 32 performs a process of searching the user dictionary 36 with the search character string Si. Then, the search command generation unit 32 performs an expansion process for extracting a character string set Yi, which is a set of character strings having the character string obtained by the search process as a representative word, from the user dictionary 36 (S350). .

  The search command generation unit 32 extracts character strings included in the character string set Xi but not included in the character string set Yi, and sets the resulting character string set as SUBi (S360). Next, the search command generation unit 32 extracts character strings that are not included in the character string set Xi but are included in the character string set Yi, and sets the resulting character string set as ADDi (S370).

  Here, each partial search command is sent from the search command generation unit 32 using the character string set SUBi and the character string set ADDi to “(IA1 like '% nA (Si)%' and not (IB like '% SUBi0%'). or IB like '% SUBi1%' or ...)) or (IB like '% ADDi0%' or IB like '% ADDi1%' or ...) ".

  Here, “IA1 like '% nA (Si)%'” indicates a process of searching the index A1 using the character string nA (Si) as a search character string. "IB like '% SUBi%'" and "IB like '% ADDi%'" are character strings that are elements constituting the character string set SUBi and the character string set ADDi. The process for searching index B for a column is shown.

  The search command generating unit 32 performs a logical operation process of the replaced partial search command (S380).

  The result of the logical operation process performed in S380 becomes a partial search command after the user dictionary expansion process. The search command generation unit 32 performs the processing up to S380 for each partial search command (S390). In this embodiment, each partial search command corresponds to a partial search command A, a partial search command B, a partial search command C, and a partial search command D.

  Then, the search command generation unit 32 performs a process of calculating the logical sum of the individual partial search commands that have been subjected to the user dictionary expansion process on the search character string, and uses the result as a search command for transmission to the search server 20. .

  Hereinafter, the user dictionary development process will be described with reference to Tables 1 and 2 with specific examples. For example, in this embodiment, the search request is a character string “computer heavy metal analysis”.

  Table 1 is an example showing a character string set as a result of extracting related character strings related to each search character string from the standard dictionary 38. Table 2 shows an example of a character string set as a result of extracting a character string related to each search character string from the user dictionary 36.

検索コマンド生成部３２は、最初に「ｓｅｌｅｃｔ＊ｆｒｏｍ ｔａｂｌｅ１ ｗｈｅｒｅ ＩＷ ｌｉｋｅ ‘％コンピュータ％’ ａｎｄ ＩＷ ｌｉｋｅ ‘％ヘビーメタル％’ ａｎｄ ＩＷ ｌｉｋｅ ‘％解析％’」という検索コマンドを生成する。この検索コマンドにおいて、部分検索コマンドとなるコマンドは、「ＩＷ ｌｉｋｅ ‘％コンピュータ％’」、「ＩＷ ｌｉｋｅ ‘％ヘビーメタル％’」および「ＩＷ ｌｉｋｅ ‘％解析％’」である。

The search command generation unit 32 first generates a search command of “select * from table1 where IW like“% computer% ”and IW like“% heavy metal% ”and IW like“% analysis% ”. In this search command, the commands that are partial search commands are “IW like“% computer% ””, “IW like“% heavy metal% ””, and “IW like“% analysis% ””.

  The partial search command generation unit 32 performs a user dictionary expansion process for each partial search command. The user dictionary expansion process in the partial search command “IW like“% computer% ”” will be described below.

  The search command generation unit 32 performs a process of searching the standard dictionary 38 using the search character string “computer”. As a result, the obtained character string nA (Si) becomes “computer”. The search command generation unit 32 extracts a character string set X, which is a set of character strings related to the character string “computer”, from the standard dictionary 38 with the character string “computer” as a representative word.

  Next, the search command generation unit 32 performs a process of searching the user dictionary 36 using the search character string “computer”. By this search process, the search command generation unit 32 obtains the character string “computer”. The search command generation unit 32 extracts a character string set Y, which is a set of character strings related to the character string “computer”, from the user dictionary 36 with the character string “computer” as a representative word.

  At this time, as shown in Table 1 and Table 2, the character string set X and the character string set Y are both composed of elements “computer, computer, electronic computer, computer”.

  In the present embodiment, when each character string set is composed of similar elements in this way, the search command generation unit 32 designates a partial search command after the user dictionary expansion process as “IA1 like '% computer%'”. And

  Similarly, the partial search command “IW like '% analysis%'” is “analysis” for both the character string set X extracted from the standard dictionary 38 and the character string set Y extracted from the user dictionary 36. , Analysis ".

  That is, since the character string set X and the character string set Y are composed of the same elements, the partial search command after the user dictionary expansion processing is “IA1 like“% analysis% ””.

  Next, a user dictionary expansion process in the partial search command “IW like“% baby metal% ”” will be described.

  First, the search command generation unit 32 performs a process of searching the standard dictionary 38 using the search character string “heavy metal”. As a result, the obtained character string nA (Si) is “heavy metal”.

  Next, the search command generation unit 32 extracts a character string set X that is a set of character strings related to the character string “heavy metal” from the standard dictionary 38 with the character string “heavy metal” as a representative word. . Here, the character string set X is composed of elements “heavy metal, heavy metal, snake meta”.

  Further, the search command generation unit 32 performs a process of searching the user dictionary 36 with the search character string “heavy metal”. As a result, the obtained character string is “tungsten alloy”. Next, the search command generation unit 32 extracts from the user dictionary 36 a character string set Y that is a set of character strings related to the character string “tungsten alloy” with the character string “tungsten alloy” as a representative word. . Here, the character string set Y is composed of elements “tungsten alloy, heavy metal, heavy metal”.

  The search command generation unit 32 extracts a character string set SUB that is included in the character string set X and not included in the character string set Y. At this time, the character string set SUB is “snake meta”.

  Next, the search command generation unit 32 extracts a character string set ADD that is not included in the character string set X but included in the character string set Y. At this time, the character string set ADD is “tungsten alloy”.

  Here, the search command generation unit 32 performs a logical operation process using the character string set SUB and the character string set ADD, and generates a partial search command after the user dictionary expansion process.

  That is, the partial search command “IW like“% baby metal% ”” becomes “(IA1 like“% baby metal% ”and not (IB like“% heavy metal% ”)) or IB like“% tungsten alloy% ””. .

  The search command generation unit 32 performs a process of calculating the logical sum of each partial search command that has been subjected to the user dictionary expansion process for each search character string, and generates a search command for transmission to be transmitted to the search server 20. .

  That is, the search command finally generated by the search command generation unit 32 is “select * from table1 where IA1 like '% computer%' and ((IA1 like '% baby metal%' and not (IB like '% heavy meta % ')) OrIB like'% tungsten alloy% ') and IA1 like'% analysis% '.

  This search command becomes a search command for sending to the search server 20 and is sent from the search client 30 to the search server 20.

  Thus, the search client 30 of the present invention can additionally register user-specific words and generates a search command using the user dictionary 36 that can be freely edited by the user. The reflected search command can be generated, and as a result, a search process customizable by the user can be executed.

  Further, the search client 30 of the present invention extracts a set of related character strings from both the standard dictionary 38 and the user dictionary 36 for each search character string, and sends a search command to be sent to the search server 20 based on the character string set. Is generated. For this reason, it is possible to perform a highly accurate search with few search omissions, and it is possible to suppress a decrease in search processing speed.

  Embodiment 2 of the present invention will be described below with reference to the drawings. FIG. 4 is an example of a functional block diagram of a search system having a search client according to the second embodiment of the present invention.

  In the search system 10A shown in FIG. 4, description of portions having the same configuration as the search system 10 described in the first embodiment is omitted. Also, in FIG. 4, the same reference numerals as those used in the description of FIG. 1 are attached to portions having the same configuration as the search system 10 described in the first embodiment.

  A search client 30A in the search system 10A shown in FIG. 4 includes a standard dictionary 38, a standard dictionary 38A, and an operation count detection unit 33. The operation count detection unit 33 detects the number of logical operation processes included in the search command generated by the search command generation unit 32.

  The search server 20A in the search system 10A performs standard normalization processing using the index A1 generated by the character string that has been subjected to standard normalization processing using the standard dictionary 38 and the standard dictionary 38A. It has an index A2 generated by the applied character string.

  Although the number of standard dictionaries held in the search client is two here, the number of standard dictionaries held in the search client is not limited, and more standard dictionaries are held. Also good.

  Further, in the search server, it is preferable that there is an index generated by a character string that has been subjected to normalization processing based on each standard dictionary corresponding to the standard dictionary held in the search client. It is not something.

  The flow of the search process in the search system 10A is the same as the flow of the search process in the search system 10 described with reference to FIG.

  The search client 20A according to the second embodiment of the present invention is different from the search client 20 according to the first embodiment in that user dictionary expansion processing is performed on a search character string using a plurality of standard dictionaries and user dictionaries.

  Hereinafter, user dictionary expansion processing according to the second embodiment of the present invention will be described with reference to FIG. FIG. 5 is a flowchart for explaining user dictionary expansion processing according to the second embodiment of the present invention.

  The search character string included in the partial search command in FIG. 5 is Si (i = 0, 1, 2,..., M), and the standard dictionary held in the search client 20A is DICp (p = 0, 1, 2, ..., m).

  The search command generation unit 32 sets S0 as the search character string included in the first partial search command among a plurality of partial search commands (S510). Next, the search command generating unit 32 determines whether the user dictionary expansion processing has been performed up to the search character string included in the last partial search command, and for the search character string that has not been subjected to the user dictionary expansion processing. The user dictionary expansion process is started (S511).

  The search command generation unit 32 sets DIC0 as the first standard dictionary used when performing the expansion processing on the search character string Si (S512), and the standard dictionary held in the search client 20A is the search character string Si. It is determined whether it is used for the user dictionary expansion process. If there is a standard dictionary that is not used in the user dictionary expansion process in the search client 20A, the user dictionary expansion process using the standard dictionary is started (S513).

  Here, the processing from S514 to S519 is the same as the processing from S330 to S370 in FIG.

  When a partial search command is generated in S519, the operation count detection unit 33 detects the number of logical operation processes included in the partial search command. Then, the search command generation unit 32 compares the number of times of the logical operation processing detected here with the minimum number of times of the logical operation processing in the number of times of the logical operation processing in the partial search commands generated so far ( S520).

  The search command generation unit 32 sets the partial search command having the smallest number of logical operation processes included in the partial search command as the partial search command in the search character string Si (S521).

  The search command generation unit 32 repeats the processing from S514 to S521 until the user dictionary expansion processing of the search character string Si is performed using all the standard dictionaries (S522).

  In this embodiment, all the standard dictionaries correspond to the standard dictionary 38 and the standard dictionary 38A. Therefore, the number of processes repeated in S522 is two. At this time, if n standard dictionaries are held in the search client, the number of processes repeated in S522 is n.

  When the search character string Si is subjected to user dictionary expansion processing using all standard dictionaries and is generated into a partial search command that minimizes the number of logical operation processes, the search command generation unit 32 The processing from S512 to S522 is repeated for the search character string included in the search command.

  In this way, when each partial search command in which the user dictionary expansion process is performed on the search character string is generated, the search command generation unit 32 performs a process of calculating the logical sum of the partial search commands, and the search server Generate search command for sending to send to.

  The second embodiment of the present invention will be described below with a specific example. In the second embodiment, as in the first embodiment, the search request is a character string “computer heavy metal analysis”.

  Tables 1 and 2 are as described in Example 1. Table 3 shows an example of a character string set as a result of extracting character strings related to each search character string from the standard dictionary 38A.

検索コマンド生成部３２は、最初に「ｓｅｌｅｃｔ＊ｆｒｏｍ ｔａｂｌｅ１ ｗｈｅｒｅ ＩＷ ｌｉｋｅ ‘％コンピュータ％’ ａｎｄ ＩＷ ｌｉｋｅ ‘％ヘビーメタル％’ ａｎｄ ＩＷ ｌｉｋｅ ‘％解析％’」という検索コマンドを生成する。

The search command generation unit 32 first generates a search command of “select * from table1 where IW like“% computer% ”and IW like“% heavy metal% ”and IW like“% analysis% ”.

  Next, the search command generation unit 32 performs a user dictionary expansion process for each partial search command. The user dictionary expansion process in the partial search command “IW like“% computer% ”” will be described below.

  The search command generation unit 32 performs a process of searching the standard dictionary 38 using the search character string “computer”. As a result, the obtained character string is “computer”. The search command generation unit 32 extracts a character string set X, which is a set of character strings related to the character string “computer”, from the standard dictionary 38 with the character string “computer” as a representative word.

  Next, the search command generation unit 32 performs a process of searching the user dictionary 36 using the search character string “computer”. By this search process, the search command generation unit 32 obtains the character string “computer”. The search command generation unit 32 extracts a character string set Y, which is a set of character strings related to the character string “computer”, from the user dictionary 36 with the character string “computer” as a representative word.

  At this time, as shown in Table 1 and Table 2, the character string set X and the character string set Y are both composed of elements “computer, computer, electronic computer, computer”.

  In the present embodiment, when each character string set is composed of similar elements in this way, the search command generation unit 32 designates a partial search command after the user dictionary expansion process as “IA1 like '% computer%'”. And

  Here, the operation count detection unit 33 detects the number of logical operation processes in the partial search command. Here, the number of logical operation processes is one.

  Next, the search command generation unit 32 performs a process of searching the standard dictionary 38A with the search character string “computer”. As a result, the obtained character string is “computer”. The search command generation unit 32 extracts, from the standard dictionary 38A, a character string set Z that is a set of character strings related to the character string “computer” with the character string “computer” as a representative word.

  At this time, the character string set Z is composed of elements “computer, computer”, and is different from the character string set Y extracted from the user dictionary 36 in that element. Therefore, here, when generating the partial search command based on the character string set Y and the character string set Z, it is necessary to perform a plurality of logical operation processes as described in the first embodiment.

  Accordingly, the search command generation unit 32 sets the partial search command related to the search character string “computer” as “IA1 like‘% computer% ’” in which the number of calculation processes is one.

  Similarly, the partial search command “IW like“% analysis% ”” includes both the character string set X and the character string set Y extracted from the standard dictionary 38 and the user dictionary 36 as elements of “analysis and analysis”. It is configured. That is, the partial search command generated here is “IA1 like“% analysis% ””, and the number of logical operation processes is one.

  Next, the search command generation unit 32 uses a character string “analysis” obtained by searching the standard dictionary 38A with the search character string “analysis” as a representative word, and is a set of character strings related to the character string “analysis”. A certain character string set Z is extracted. At this time, the character string set Z and the character string set Y are both composed of elements of “analysis and analysis”. That is, the number of logical operation processes of the partial search command generated here is also one.

  Therefore, the search command generation unit 32 sets the partial search command related to the search character string “analysis” to “IA1 like“% analysis% ””.

  Next, the user dictionary expansion process of the partial search command “IW like“% heavy metal% ”” will be described below.

  As described in the first embodiment, the partial search command generated when the user dictionary expansion process is performed based on the standard dictionary 38 and the user dictionary 36 is “(IA1 like“% baby metal% ”and not (IB like). '% Heavy metal%')) orIB like '% tungsten alloy%'.

  Next, the search command generation unit 32 performs a process of searching the standard dictionary 38A with the search character string “heavy metal”. As a result, the obtained character string becomes “tungsten alloy”. The search command generation unit 32 extracts, from the standard dictionary 38A, a character string set Z that is a set of character strings related to the character string “tungsten alloy” with the character string “tungsten alloy” as a representative word. At this time, the character string set Z is composed of elements “tungsten alloy, heavy metal, heavy metal”, and is the same as the character string set Y in the elements.

  That is, the partial search command generated based on the character string set Y and the character string set Z is “IA2 like“% heavy metal% ””, and the number of logical operation processes is one.

  Therefore, the search command generation unit 32 sets the partial search command related to the search character string “heavy metal” to “IA2 like“% heavy metal% ””. When the partial search command is generated in this way, the search command generation unit 32 performs a process of taking a logical product of these, and generates a search command for sending to the search server 20A.

  Here, the search command sent to the search server 20A is “select * from table1 where IA1 like“% computer% ”and IA2 like“% heavy metal% ”and IA1 like“% analysis% ”.

  As described above, the search client 30A according to the second embodiment generates a partial search command with the smallest number of calculation processes and sends it to the search server 20A. As a result, the load on the search server 20A is reduced, and faster search processing can be performed.

  The standard dictionary 38 and the standard dictionary 38A described in the first and second embodiments of the present invention can be updated by the standard dictionary updating unit 34. That is, when there is a change in the standard dictionaries held in the search client 20 and the search client 20A, standard dictionary update data, which is changed standard dictionary information, is distributed from the standard dictionary distribution unit 28. The standard dictionary update unit 34 updates the standard dictionary 38 and the standard dictionary 38A based on the standard dictionary update data.

  As described above, according to the present invention, it is possible to provide a search client that can execute high-precision search processing with few search omissions and can suppress a decrease in search processing speed.

  Also, to provide a search client capable of executing a search process customized on the user side by reflecting a user-specific characteristic by generating a search command using a user dictionary that can be freely edited by the user. Can do.

  Further, by generating a search command with the smallest number of logical operation processes, it is possible to provide a search client that can reduce the load on the search server and suppress a decrease in search processing speed.

  Furthermore, by updating the standard dictionary held in the search client, it is possible to always execute a search process based on the latest information, and to continue other effects of the present invention. Can be provided.

  Although the present invention has been described based on each embodiment, the present invention is not limited to the requirements shown in the above embodiment. With respect to these points, the present invention can be changed within a range that does not detract from the gist of the present invention, and can be appropriately determined according to the application form.

  The present invention can be applied to a search client constituting a search system that performs search processing.

It is a functional block diagram of the search system which has a search client in Example 1 of this invention. 3 is a flowchart showing a flow of search processing in the search system 10. It is a flowchart explaining the user dictionary expansion | deployment process in Example 1 of this invention. It is a functional block diagram of the search system which has a search client in Example 2 of this invention. It is a flowchart explaining the user dictionary expansion | deployment process in Example 2 of this invention.

Explanation of symbols

10, 10A Search system 20, 20A Search server 21 Standard dictionary distribution unit 30, 30A Search client 32 Search command generation unit 33 Operation count detection unit 34 Standard dictionary update unit 36 User dictionary 38, 38A Standard dictionary 39 Character normalization dictionary A1, A2, B Index

Claims (2)

N words corresponding to the n standard dictionaries generated by the standardized character strings using n standard dictionaries that collect the basic word data and the normalization results when the language is analyzed. A standard index,
A search server having a character unit normalization index generated by a character string normalized using a character normalization dictionary for performing character unit normalization processing on a search character string included in a search request In a search client that accesses and searches,
The search client
Editable user dictionaries that can register additional user-specific words and collect special word data for each individual,
The n standard dictionaries;
The character normalization dictionary;
Search command generation means for generating a search command;
An operation number detecting means for detecting the operation number of the logical operation included in the search command ,
The search command generation means includes
The search character string is independently searched in one standard dictionary and the user dictionary among the n standard dictionaries,
A set der Ru collection if Xi of the string to the representative words a string of search results of the one of the standard dictionary, set der Ru collection of string to a representative word a string of search results in the user dictionary Extract Yi ,
A first set of results obtained by searching the standard index corresponding to the one standard dictionary using the elements of the set Xi as a search character string;
A set obtained by searching the character unit normalized index with each element constituting the set SUBi included in the set Xi and not included in the set Yi, and a set included in the set Yi and not included in the set Xi A second set of results obtained by ORing with a set of results obtained by searching the character unit normalized index with each element constituting ADDi;
A search command including a logical operation for extracting the first set that is not included in the logical product of :
The search client
In the search command generating means, the search command having the smallest number of arithmetic operations detected by the arithmetic operation number detecting means among the search commands generated corresponding to each of the n standard dictionaries is searched. A search client characterized by being sent to a server.   The search client according to claim 1, further comprising a standard dictionary update unit configured to update the standard dictionary based on standard dictionary update data distributed from the search server.

Images