JP2002132809A - Character string retrieval method, its implementation device, and recording medium recorded with processing program thereof - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide technology capable of improving the usability of character string retrieval. SOLUTION: The character string retrieval method for retrieving a character string has a step for extracting syllable elements from an inputted keyword, a step for normalizing representation vacillation of the extracted syllable elements, and a step for retrieving a character string matching the reading of the inputted keyword by referring reading matching indexes indicating storage locations of character strings including the normalized syllable elements.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character string search apparatus for searching a character string, and more particularly to a character string search apparatus for searching a Japanese character string written in Roman characters in a directory service based on its reading. And effective technology.

[0002]

2. Description of the Related Art As a means for realizing smooth communication within a company and between companies, a LAN (Local Area Net) is used.
work) and other PCs (Personal Compu
e-mail systems for transmitting and receiving documents created by information processing devices such as ter.) have been widely used, and as a means for searching for a recipient's e-mail address, a so-called electronic address book function, the X.T. Directory services, such as 500 (ISO9594), are used.

[0003] X. The directory service conforming to 500 has a data model hierarchically managed as a tree structure. Directory entries are arranged at locations corresponding to the branches and leaves of the tree. Each entry has a name (DN:
Distinguished Name), and in addition to the user's e-mail address, various information such as first and last name, telephone number, FAX number, and photograph can be stored as attributes.

X. Reference numeral 500 adopts a client-server type distributed system architecture, and a DAP (Directory Access Protocol) according to a seven-layer structure of OSI (Open Systems Interconnection) as a communication protocol between information processing apparatuses that play a role of a client and a server.
col). On the other hand, IETF (Internet Engineering Tas
k Force) is "LDAP: Lightwei as a directory client-server protocol on TCP / IP.
ght Directory Access Protocol (RFC2251) "
Was standardized. The user can send an X.X. A directory server such as 500 is accessed by DAP or LDAP to search for desired information such as a user's mail address. DAP or LDAP also defines directory update requests such as entry addition, deletion, update, and entry name change.

X. One of the search functions specified in the 500 is an attribute search function. The attribute search function is a function of searching for a directory entry that matches the attribute value specified by the user. For example, the attribute search for the user's first and last names is a function corresponding to a so-called Japanese syllabary address book. The attribute search function includes an all-match search for searching for directory entries that all match the attribute value specified by the user, and a partial-match search for searching for directory entries that partially include them.

[0006] The attribute search can be realized by a sequential search method used in various information systems. But,
In the sequential search method, attribute values are compared for all entries stored by the directory service, so that the processing overhead is large and sufficient performance cannot be obtained. In particular, at the time of a partial match search, the performance is significantly deteriorated.

[0007] In order to solve the above problem at the time of partial match search, for example, in 1995, the URL ftp://terminator.rs.
Ti published on itd.umich.edu/ldap/papers/xldbm.ps
Mothy A Howes's document "An X.500 and LD
“4.3 Substring Ma” described on pages 4 and 5 of “AP Database: Design and Implementation”
A method of “tching” and an index search method as disclosed in JP-A-7-319920 have been proposed.

FIG. 21 is a diagram showing an example of the information structure of a directory DB in which a conventional directory service stores each directory entry. In FIG. 21, rows are entries and columns are attributes. Directory DB configuration example 1501
Indicates the attributes of each user entry as DN (dn), English surname (sn), English name (givenName), and email address (mai
l), telephone numbers (tel), and fax numbers (fax). To facilitate processing, each directory entry has an entry ID (ID) which is a unique identification number.
Is added.

FIG. 22 is a diagram showing an example of an information structure of a pattern matching index for an English name attribute in a conventional index search method. In the conventional index search method, before referring to the directory DB configuration example 1501,
By narrowing down candidates using the pattern match index table 1601 in advance, the speed of the partial match search process is increased.

As shown in FIG. 22, the pattern matching index table 1601 includes a character string storage area for storing a pattern matching key 1602 and a list storage area for storing an entry ID list 1603. Entry ID list 160
The entry ID registered in No. 3 corresponds to the entry ID of each directory entry shown in FIG. The pattern matching key 1602 is a partial character string obtained by extracting a total of n characters (m is 0 and n is 3 in the figure) one by one every m characters from the character string that is the attribute value.

For example, if the DN is "cn = Hiroshi Sato, ou = Sale
At the time of registration of the entry of “s, o = Hitachi, c = JP”, after registering each attribute value in the directory DB configuration example 1501, “H
For the English name "iroshi", "HIR", "IRO", "RO
Five search keys S "," OSH ", and" SHI "are generated, and together with the corresponding entry ID" 1 ", the pattern match key 1602 and the entry ID in the English name pattern match index table 1601 shown in FIG. It is registered in the list 1603. A pattern matching index table 1601 is similarly generated for other attributes.

On the other hand, at the time of retrieval, similarly, n characters are extracted every m characters from the retrieval character string, and a corresponding entry ID list 1603 is extracted from the pattern matching index table 1601 using each as a retrieval key. The candidate is narrowed down by selecting an entry ID common to the entry ID list 1603 of. For example, the English name "hir
"HI" to search for entries containing the string "o"
Two search keys R "and" IRO "are generated, and an entry ID corresponding to each pattern matching key 1602 is obtained from the English name pattern matching index table 1601 shown in FIG.
The list 1603 is extracted. Next, the two extracted entry ID lists 1603, "1,2,4,5,6" and "1,2,5,6" are compared, and a common entry ID "1,2,5" , 6 "as a search result. That is, the English names are "Hiroshi", "Shiro", "C
The search result is an entry that is "hihiro" or "Ichiro".

[0013]

According to the above-described conventional index search method, the following problem occurs when a Japanese user searches for an attribute registered in Roman characters. Taking the case of FIG. 22 as an example, when a Japanese user requests the directory service to search for an entry containing the character string “hiro”, the user entry whose name is “Hiro” is usually desired. it seems to do. However, according to the conventional index search method, since all entries including the search character string are used as search results using a character string pattern matching method, "Hiroshi", "Chihiro" In addition to "", the search results include entries that do not include the English names "Shiro" and "Ichiro" that the user does not want. In addition, there is fluctuation in the Roman alphabet notation depending on the user. For example, in FIG.
In the entry whose DN is "cn = Satiko Suzuki, ou = Sales, o = Hitachi, c = JP", "Satiko" is registered as an English name attribute. In such cases, another Romanized notation, "Sac
Even if the English name is searched by "hiko", the above entry does not hit. An object of the present invention is to provide a technique capable of solving the above problem and improving the usability of the character string search.

[0014]

SUMMARY OF THE INVENTION The present invention provides a character string search apparatus for searching for a character string based on the reading of a keyword. In the present invention, when registering a character string to be searched in a database, a syllable element is extracted from the character string, and a plurality of different notations existing for the reading of the extracted syllable element are unified. Performs normalization of the notation fluctuation to be converted to. Then, the normalized syllable element and the storage location of the character string are associated with each other and stored as a reading matching index.

When a user inputs a keyword when searching for a character string, a syllable element is extracted from the input keyword, and the notation fluctuation of the extracted syllable element is normalized in the same manner as described above. Then, by referring to the reading matching index, the character string is read from the storage location of the character string including the syllable element of the normalized keyword, and a character string matching the reading of the input keyword is searched.

For example, when searching attribute information indicating an English name in which a Japanese name is expressed in Roman characters by reading Japanese from a directory DB storing various attribute information provided by a directory service, first, After extracting and normalizing the syllable elements when the English name of the Roman alphabet stored in the directory DB is read in Japanese, the directory DB of each attribute information including the normalized syllable elements and the alphabetic names is obtained. Is stored in the form of a reading match index in association with the entry ID.

When performing a search for an English name in which a Japanese name is expressed in Roman characters, the English name input by the user as a keyword is decomposed into syllable elements when read in Japanese in the same manner as described above, and a regular expression is obtained. After the conversion, the entry ID corresponding to the normalized syllable element is read from the reading match index, and the directory entry of the directory DB is accessed to search for an English name that matches the input keyword.

As described above, according to the present invention, it is possible to remove a search result different from the user's intended reading by using a reading matching index based on the reading of a character string, and to obtain a desired search result even if there is fluctuation in the notation. Thus, it is possible to provide a directory search method that is easy to use for the user. As described above, according to the character string search device of the present invention, since the character string is searched based on the reading of the keyword, the usability of the character string search can be improved.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS (Embodiment 1) A character string search apparatus according to Embodiment 1 for searching a Japanese character string written in Roman characters in a directory service based on its reading will be described below. The directory service of the present embodiment realizes a search function based on reading by decomposing each attribute information into one or more syllable elements and using a reading match index associated with a directory entry of the attribute information. . In the present embodiment, the comparison of the reading matches is a comparison for checking whether the character strings to be compared have the same pronunciation. For example, in Japanese in Roman notation, “Hiroshi” and “Hirosi” do not match in a comparison of character strings, but have the same reading, so that in a comparison of reading matching, they are regarded as a match.

FIG. 1 is a diagram showing a schematic configuration of a system for providing a directory service according to the present embodiment. As shown in FIG. 1, in the directory service of the present embodiment, LA
Directory server 101 on the network 103 such as N
And the client 102 are connected.

The directory server 101 stores a directory DB 104 for storing various types of attribute information on directory entries, index definition information 105 for storing types of indexes to be assigned to each attribute information of the directory entries, and stores one or more pieces of each attribute information. Pattern matching index 10 to be decomposed into pattern elements and stored
6. a reading matching index 107 for decomposing each attribute information into one or more syllable elements and storing the directory matching information;
It comprises a reading matching key extracting unit 109 for extracting a reading matching key of the reading matching index 107 from attribute information and the like, and a pattern matching key extracting unit 110 for extracting a pattern matching key of the pattern matching index 106.

The directory control unit 108 includes a directory update control unit 111 for controlling a directory update process, a directory search control unit 112 for controlling a directory search process, an index update control unit 113 for controlling an index update, and an index search. Is controlled by an index search control unit 114.

The index update control unit 113 includes a pattern match index update unit 115 for executing a write process on the pattern match index 106, and a read match index update unit 116 for executing a write process on the read match index 107. .

The index search control unit 114 includes a pattern match index search unit 117 that executes a search process on the pattern match index 106 and a reading match index search unit 118 that executes a search process on the reading match index 107. .

The reading matching key extracting unit 109 extracts a syllable element from the input keyword or attribute information of the directory entry, etc., and a syllable element normalizing unit that normalizes the fluctuation of the notation of the extracted syllable element. Part 1
20. The pattern matching key extracting unit 110 includes a key element extracting unit 121 that extracts an element serving as a pattern matching key of the pattern matching index 106 from attribute information and the like.
It comprises a pattern normalizing section 122 for normalizing the pattern.

In this embodiment, the information elements of the pattern matching index 106 and the pattern matching index updating unit 115, the pattern matching index searching unit 117,
Pattern matching key extraction unit 110, key element extraction unit 12
1. The processing contents of the pattern normalization unit 122 are the same as in the conventional example.

A program for causing the directory server 101 to function as each processing unit is recorded on a recording medium such as a CD-ROM, stored on a magnetic disk or the like of the directory server 101, and then loaded into a memory of the directory server 101. Shall be executed. The recording medium for recording the program may be a recording medium other than the CD-ROM.

FIG. 2 is a diagram showing an example of the information structure of the reading coincidence index 107 according to this embodiment. In FIG. 2, FIG.
Matching Index Table 2001 for English Name Attributes
And a reading match key 200 serving as a search key as in the case of the conventional pattern match index 106.
2 and an entry ID list 2003. However, the reading match key 20 in the reading match index table 2001
02 is a partial character string obtained by decomposing a character string as an attribute value into syllable elements. First, an index generation method will be described. The generation of the index is performed not only at the time of entry registration but also at the initial construction of the directory DB 104. When an entry or attribute value is deleted, the index is deleted, and when the attribute value is changed, the index is changed.

FIG. 3 is a diagram showing an example of the information configuration of the index definition information 105 of the present embodiment. FIG. 3 shows an example of the information configuration of the index definition information 105 that stores the type of index given to each attribute of the directory entry by the operation manager of the directory server 101. The index definition information 1701 in FIG. 3 includes a storage area for attribute names and three areas for storing necessity / unnecessity for each index type. Directory server 101
The operation administrator of the site must know which attributes need to be indexed,
In the index type area to be created, “Y” indicating creation is registered, and if index creation is not required, “N” indicating index not needed is registered.

FIG. 4 shows a directory control unit 1 according to this embodiment.
It is a flowchart which shows the processing procedure of 08. The directory control unit 108 of the directory server 101 reads a processing request from the directory client 102 (step 201), determines the content of the processing request (steps 202 to 206), updates the directory (step 208), or searches for a directory (step 201). 20
9), and returns a processing result to the directory client 102 (step 207). When the processing request from the directory client 102 is an entry registration request, an entry change request, or an entry deletion request, the directory update control unit 111 is called (step 20).
8).

FIG. 5 is a flowchart showing a processing procedure of the directory update control unit 111 of this embodiment. When the processing request from the directory client 102 is a request for registering a directory entry, the directory update control unit 111 called from the directory control unit 108 determines the entry identifier, the English surname,
Attributes such as an English name, a mail address, a telephone number, and a FAX number are registered in each storage area in the directory DB 104 (step 301). Similarly, in the case of an entry change request, each storage area in the directory DB 104 is changed according to the change information included in the request. The same applies to a deletion request. Next, the directory update control unit 111 calls the index update control unit 113 to update the index update of each attribute (step 3).
02). This call has as parameters the name of the attribute, a character string that is the attribute value, and an entry ID that can uniquely identify the directory entry.

FIG. 6 is a flowchart showing the processing procedure of the index update control unit 113 of this embodiment. The index update control unit 113 called from the directory update control unit 111 searches the index definition information 105 for a record having the same attribute name as the attribute name specified to the directory update control unit 111 (Step 501). If “Y” is set to the entire matching area or the partial matching area in the searched record (step 502), the pattern matching index updating unit 115 is instructed to update the pattern matching index (step 505). This instruction has, as parameters, the type of index to be updated (all match or partial match), a character string as an attribute value, and an entry ID. Further, when “Y” is set in the reading matching area in the searched record (step 503), the reading matching index updating unit 116 is instructed to update the reading matching index (step 506). This instruction has a character string as an attribute value and an entry ID as parameters. The processing from step 501 is repeated until all the update information is processed (step 504).

FIG. 7 is a flowchart showing a processing procedure of the pattern matching index updating unit 115 of the present embodiment. The pattern matching index updating unit 115 called from the index updating control unit 113 calls the pattern matching key extracting unit 110, and the pattern matching key 160
2 is extracted (step 701). If the instruction is a registration request (step 702), the pattern matching index table 1
The entry ID is registered in the entry ID list 1603 associated with each of the extracted pattern matching keys 1602 at step 601 (step 706). If the instruction is a deletion request (step 703), the entry ID is deleted from the entry ID list 1603 associated with each extracted pattern matching key 1602 in the pattern matching index table 1601 (step 707). If the instruction is a change request (step 704), the entry ID is deleted from the entry ID list 1603 associated with each pattern matching key 1602 extracted from the pre-change information in the pattern matching index table 1601 (step 708). The entry ID is registered in the entry ID list 1603 of the pattern matching index table 1601 associated with each pattern matching key 1602 extracted from the post-change information (step 709). The process is repeated from step 701 until all update information is processed (step 705).

FIG. 8 is a flowchart showing a processing procedure of the pattern matching key extraction unit 110 of this embodiment. The pattern matching key extracting unit 110 called by the pattern matching index updating unit 115 first normalizes the fluctuation of the notation of the given pattern by the pattern normalizing unit 122 (step 1101). Next, the key element extraction unit 121 extracts n characters from the given pattern at every m characters (step 1102). The process is repeated from step 1102 until there is no more key element (step 1103). The pattern matching key extraction unit 110
Return the extracted key to the caller.

FIG. 9 is a flowchart showing a processing procedure of the reading matching index updating unit 116 according to the present embodiment.
The reading match index updating unit 116 of the present embodiment extracts the syllable element by the syllable element extracting unit 119 and
The registration, deletion or change of the entry ID of the attribute information including the syllable element is performed for each of the syllable elements normalized by the above, and the matching index table 2001 is read. The processing procedure of the reading matching index updating unit 116 of the directory server 101 is the same as the processing of the pattern matching index updating unit 115 shown in FIG. 7, but the key is extracted by using the reading matching key extracting unit 109 (step). 801) Different. That is, the index update control unit 11
Reading match index updating unit 116 called from step 3
Calls the reading matching key extraction unit 109 and extracts the reading matching key 2002 (step 801). If the instruction is a registration request (step 802), the entry ID is registered in the entry ID list 2003 associated with each extracted reading matching key 2002 in the reading matching index table 2001 (step 806). If the instruction is a deletion request (step 803), the reading match index table 2
001, the entry ID is deleted from the entry ID list 2003 associated with each extracted matching key 2002 (step 807). If the instruction is a change request (step 804), the reading match index table 20
01, each reading matching key 200 extracted from the pre-change information
2 from the entry ID list 2003 associated with
The entry ID is deleted (step 808), and the entry ID associated with each reading matching key 2002 extracted from the post-change information in the reading matching index table 2001
The entry ID is registered in the list 2003 (step 809). The process is repeated from step 801 until all update information is processed (step 805).

FIG. 10 is a flowchart showing a processing procedure of the reading matching key extraction unit 109 of the present embodiment. The reading matching key extracting unit 109 called from the reading matching index updating unit 116 refers to the Roman alphabet table by the syllable element extracting unit 119 and recognizes the first syllable element of the given character string (step 1201). FIG. 11 is a diagram showing an example of the Roman alphabet notation table of the present embodiment. Although omitted in the drawing, it is assumed that there is also a table in Roman alphabet notation such as a muddy sound, a plosive sound, and a murmur, and a table indicating syllables in different notations for the same reading. The syllable element extraction unit 119 of the present embodiment compares the syllables in the Roman alphabet notation table 1801 with the character string that is the attribute value, and extracts those that match the syllables in the Roman alphabet notation table 1801 as syllable elements of the character string. I do. Next, the matching key extraction unit 109 of the directory server 101
The syllable element normalization unit 120 normalizes the recognized syllable elements with reference to the Roman character normalization table (step 1202).

FIG. 12 is a diagram showing an example of the Roman character normalization table of the present embodiment. The syllable element normalizing section 120 of the present embodiment compares the syllable element extracted by the syllable element extracting section 119 with the area 19 in the Roman character normalization table 1901.
The syllable element extracted in the syllable element extraction unit 119 is compared with the notation registered before normalization registered in the area 1902, and if the syllable element is the notation registered in the area 1902, the syllable element is stored in the area 1902.
3 is converted into a notation of 3, and a plurality of different notations existing for the reading of the syllable element are converted into a unified notation, and the notation fluctuation is normalized. For example, when the syllable element recognized by the syllable element extraction unit 119 is “HU”, the expression for the pronunciation of “fu” is unified and normalized by converting the syllable element to “FU”.

Step 1201 is repeated until there is no character string following the recognized syllable element. If the string runs out, it returns the recognized and normalized syllable element. It should be noted that the Roman character notation table 1801 and the Roman character normalization table 1901 are provided in the reading matching key extraction unit 109.

For example, when the given attribute value is “Hiroshi”, the reading matching key extraction unit 109 refers to the Roman alphabet notation table 1801 and detects the syllable element “Hi” of the character string. The syllable-decomposed partial character string "Hi" has no fluctuation in notation, but if the syllable-decomposed partial character string has fluctuation in notation, then the reading matching key extraction unit 109
Corrects fluctuations in the notation in the Roman character normalization table 1901 and also corrects fluctuations in uppercase and lowercase letters. As a result of such processing, "HI", "RO", and "SHI" of the read matching key 2002 are obtained from the attribute value "Hiroshi". Similarly, "Sati
From "ko", "SA", "CHI", and "KO" are obtained, and these reading match keys 2002 are used by the reading match index updating unit 116 shown in FIG.

Next, an index search method will be described. FIG. 13 is a diagram showing a screen display example at the time of attribute search according to the present embodiment. FIG. 13 illustrates an example of a screen display when the end user searches the directory, and the directory client 102 controls display on the display and input using a keyboard, a mouse, or the like.

The screen 1301 is an area 1 for selecting an attribute name.
302, an area 1303 indicating a search pattern character string,
Area 130 for selecting reading match search or pattern match search
4. An area 1305 for selecting a partial match or a perfect match, an OK button 1306 for executing a search under set conditions, and a cancel button 13 for canceling the set information
07. In the present embodiment, if the area 1305 for selecting a partial match or a perfect match is “contains”, it is an arbitrary character string before and after the character string input to the area 1303 indicating the search pattern character string. It is assumed that a wildcard "*" indicating that the event is to be performed is specified.

When various information is input on a screen 1301 and an OK button 1306 is pressed, the directory client 102 sends an LDA to the directory server 101.
Issues a search request such as P. As described in the flowchart of FIG. 4, the directory control unit 108 of the directory server 101 that has received the search request from the directory client 102 recognizes that the request is an entry search (step 206), and Call 112 (step 209).

FIG. 14 is a flowchart showing a processing procedure of the directory search control unit 112 according to this embodiment. Upon receiving the search condition from the directory control unit 108, the directory search control unit 112 passes the search condition, calls the index search control unit 114, and obtains a list of IDs of hit entry candidates (step 40).
1). While there is a search hit candidate (step 402), entry information is read from the directory DB 104 using the entry ID as a key (step 403). Then, it is checked whether the read entry information matches the search condition (step 404). If they match, the entry information is returned to the directory client 102 (step 405). If they do not match, the next hit candidate is checked (step 406).

FIG. 15 is a diagram showing an example of a screen display of an attribute search result according to the present embodiment. Upon receiving the search result, the directory client 102 displays a screen 1401 as shown in FIG.
Output the search result to.

FIG. 16 is a flowchart showing the processing procedure of the index search control unit 114 of this embodiment. When the given search condition is a pattern match (step 601), the index search control unit 114 called from the directory search control unit 112 calls the pattern match index search unit 117 (step 60).
6) If the given search condition is a reading match (step 602), the reading matching index search unit 118
To obtain a list of entry IDs of search hit candidates under the condition. Processing is continued from step 601 until all search conditions are processed (step 603). After processing all search conditions, the list of entry IDs of hit candidates found in each search condition is
A logical operation is performed according to a logical operation condition between the search conditions (step 604). A list of entry IDs resulting from the logical operation is returned to the caller (step 605).

FIG. 17 is a flowchart showing a processing procedure of the pattern matching index search unit 117 of this embodiment. The pattern match index search unit 117 called by the index search control unit 114 uses the pattern match key
Is obtained (step 901). Step 90
In step 2, the entry ID list 1603 associated with the acquired pattern matching key 1602 is read from the pattern matching index table 1601, and step 901 is executed.
The processing of step 902 is repeated until all the pattern matching keys 1602 extracted in step are processed (step 90).
3). Then, each of the read entry ID lists 16
The logical product of the numbers 03 is obtained and returned as an index search result (step 904).

FIG. 18 is a flowchart showing a processing procedure of the reading matching index search unit 118 of this embodiment. As shown in FIG. 18, the reading matching index search unit 118 of the directory server 101
For each syllable element of the keyword extracted in step 9 and normalized by the syllable element normalizing unit 120, the reading matching index table 2001 is referenced to search for attribute information including those syllable elements. The reading matching index searching unit 118 called from the index searching control unit 114 calls the reading matching key extracting unit 109 and requests a list of reading matching keys 2002 (step 1001).

The reading matching key extracting unit 109 called from the reading matching index searching unit 118 performs the same processing as that of the reading matching index updating unit 116, and the syllable element extracting unit 119 causes the Roman alphabet notation table 180.
1, the first syllable element of the character string given as the keyword for the search is recognized (step 1201). The syllable element extraction unit 119 uses the Roman alphabet notation table 1801
The middle syllable and the character string of the keyword are compared, and a syllable that matches the syllable in the Roman alphabet notation table 1801 is extracted as a syllable element of the character string.

Next, the reading matching key extracting section 109 normalizes the recognized syllable element by referring to the Roman character normalization table by the syllable element normalizing section 120 (step 120).
2). The syllable element normalizing section 120 includes the syllable element extracting section 11
9 is compared with the notation before normalization registered in the area 1902 in the Roman character normalization table 1901, and the syllable element of the keyword is set in the area 1
If the notation is registered in 902, the syllable element is converted to the notation in the area 1903 and normalized. Step 1201 until there is no character string following the recognized syllable element.
It repeats from and, when the character string runs out, recognizes,
The normalized syllable elements are read and returned as a list of matching keys 2002.

The reading match index search unit 118 called from the index search control unit 114 acquires the list of the reading matching keys 2002 returned from the reading matching key extraction unit 109 (step 1001). Next, in step 1002, the entry ID list 2003 associated with the acquired matching key 2002 is read from the reading matching index table 2001, and step 1001 is executed.
Step 1002 is repeated until all the read matching keys 2002 extracted in Step 100 are processed (Step 100).
3). Then, each of the read entry ID lists 20
The logical product between the numbers 03 is obtained and returned as an index search result (step 1004).

In the example of FIG. 2, by calling the reading matching key extraction unit 109 in step 1001, the search condition "hir"
“HI” and “RO” are extracted from “o”.
HI "is used as a key to search for a match key 2002 to obtain entry IDs" 1,5 "from the entry ID list 2003. Similarly, entry IDs" 1,2,4,5,6 "are obtained from" RO ". In step 1003, the logical product of the two is "1,5", and in step 1004, "1,5" is returned as the index search result.

The directory search method according to the present embodiment has been described. According to the conventional index search method, since all entries including the search character string are set as search results, the search results include entries not desired by the user. On the other hand, the directory search method according to the present embodiment can obtain a search result faithful to the reading of the attribute value by using the reading matching index in which the attribute value is decomposed into syllable elements. For example, according to FIG.
Interpret the partial match search request for "o" as a partial match search request for the attribute value of the reading "Hiro", and register "Shiro" or "Ichiro" that includes "hiro" but does not include the reading "Hiro""Hiroshi" and "Chih"
Only entries with "iro" registered can be used as search results.

Further, according to the present embodiment, the user can search for a desired entry even if the notation of the attribute value fluctuates. For example, according to FIG. 2, a partial match search request for “sachi” is interpreted as a partial match search request for an attribute value of “sachi”, and “sachi” is not included but contains “sati”, which is a synonym. The entry can be a search result. As described above, according to the character string search device of the present embodiment, the character string is searched based on the reading of the keyword, so that the usability of the character string search can be improved.

(Embodiment 2) The following is a second embodiment in which a directory service performs a full match character string search of a Japanese character string represented in Roman characters in consideration of its fluctuation.
Will be described. Embodiment 1 above
Relates to a method of searching for a partial character string based on the reading. However, the directory search method of the present invention also takes into account the fluctuation of the notation, and is applicable to an all-match character string search.

FIG. 19 is a diagram showing an example of the information structure of the reading coincidence index 107 according to this embodiment. FIG. 19 shows an example of the information configuration of a reading match index table 2101 that enables a full match and partial match search based on reading.
The reading match index updating unit 116 of the present embodiment extracts the syllable element by the syllable element extracting unit 119 and
The registration, deletion or change of the entry ID of the normalized attribute information obtained by concatenating the syllable elements with respect to each syllable element is performed on the reading match index table 2101, and the reading match index search unit 118 Is
The syllable element extracting unit 119 extracts the syllable element normalizing unit 12
For each syllable element of the keyword normalized by 0, a full match character string search is performed by referring to the reading matching index table 2101 and searching for attribute information matching the normalized keyword obtained by connecting the syllable elements. Other configurations are the same as in the first embodiment.

The reading match index table 2101 in FIG.
In the same manner as in the first embodiment (FIG. 2), a character string obtained by dividing an attribute value for each syllable is registered, and the syllable elements normalized by the reading matching key extraction unit 109 are sequentially connected, that is, A normalized value of the entire attribute value is registered. When this embodiment is applied, in the reading matching index updating unit 116 shown in FIG. 9, after the reading matching key is extracted (step 801), the normalized syllable elements are sequentially connected, that is, the entire attribute value is normalized. What is done may be added to the reading matching key 2102.

Further, in the index search operation of the reading match index search unit 118 shown in FIG. 18, if the search type is a full match search based on reading, after the reading matching key extraction (step 1001), the normalized syllable element May be added to search for the reading matching key 2102 of the reading matching index table 2101 using a sequence obtained by normalizing the entire attribute value, and a corresponding entry ID list 2103 may be extracted and returned. .

Taking FIG. 19 as an example, the English name attribute "Satiko"
When an index is generated, the reading match index updating unit 116 performs a full match search index “SACHIKO” and a partial match search index “S
A "," CHI ", and" KO "are read and the match index table 2101 is read.
Register with. At the time of an all-match search based on readings, the reading matching index search unit 118 uses the attribute value “Sachi
After normalizing “ko”, an entry ID “3” corresponding to “SACHIKO” is found and set as a search result.

As described above, according to the present embodiment, it is possible to obtain an all-match search result faithful to reading attribute values. For example, according to FIG. 19, an all-match search request for "sachiko" is interpreted as an all-match search request for the attribute value of the reading "Sachiko".
The entry "ko" can be used as a search result. As described above, according to the character string search device of the present embodiment, a search for a character string is performed based on the reading of a keyword, thereby improving the usability of the character string search. It is possible.

(Embodiment 3) A character string search for a Japanese character string described in Roman characters in a directory service is performed in consideration of the continuity of its syllables.
Will be described. Embodiment 1 above
In, attribute values are separated by syllables and indexed for each syllable. However, this method does not consider the continuity of syllables. For example, if the English name attribute "Hikoroku" is added to the reading match index table 2101 in FIG. 2, the entry will also be hit at the time of a partial match search for "hiro". . This problem can be solved by a process of narrowing down candidates by index search, reading information of each entry from the directory DB 104, and checking again whether the search condition is met (step 404). Refining as much as possible can improve search performance.

In order to solve the above problem, in the directory search method according to the present embodiment, after the attribute value is decomposed into syllable elements, a process of indexing a character string corresponding to n syllables every m syllables is performed. In the present embodiment, m is 1,
Let n be 2.

FIG. 20 is a diagram showing an example of the information structure of the reading coincidence index 107 according to this embodiment. As in the first embodiment (FIG. 2), a character string obtained by dividing the attribute value for each syllable is registered in the reading matching index table 2201 of FIG. 20, and the attribute value is extracted for every two syllables. The registered character string is registered.

The reading matching index updating section 116 of this embodiment extracts a plurality of syllable elements extracted by the syllable element extracting section 119 and normalized by the syllable element The entry ID of the attribute information including the syllable element is registered, deleted, or changed in the reading matching index table 2201, and the reading matching index searching unit 118 extracts the syllable element extracting unit 119, and the syllable element normalizing unit. For each syllable element of the keyword normalized in 120, the reading match index table 220
1 and search for attribute information including a plurality of continuous syllable elements among those syllable elements. Other configurations are the same as in the first embodiment.

When this embodiment is applied, in the index updating operation of the reading matching index updating unit 116 shown in FIG. 9, the character strings divided for each syllable are sequentially registered in the reading matching index table 2201, and then, A character string extracted every two syllables at every 0th syllable may be sequentially registered in the reading match index table 2201 as the reading match key 2202.

Further, in the index search operation of the reading matching index search unit 118 shown in FIG. 18, if the attribute value in the search condition is composed of a plurality of syllables, the character string is divided into two syllables every 0 syllable, and Entry ID list 2 corresponding to later two-syllable character string reading match key 2202
After extracting 203, processing may be added to return a common entry ID.

Taking FIG. 20 as an example, the English name attribute “Hikorok
When generating an index for "u", the reading matching index update unit 116 adds "HI", "KO", "RO", and "KU" for one syllable and "HIKO" and "KORO" for every two syllables. , "ROKU",
It is registered in the reading matching index table 2201. When searching for a plurality of syllables, the reading matching index search unit 118 divides the attribute value “hiro” in the search condition into two syllables at intervals of 0 syllable, and the entry ID “1,5” corresponding to “HIRO” Is found and set as a search result. In the case of a partial match search for "hiroshi", similarly, it is divided into two syllables at intervals of 0 syllable, and an entry ID "1" corresponding to both "HIRO" and "ROSHI" is found. And

As described above, according to the present embodiment, it is possible to perform a search process in consideration of the continuity of syllables. For example, according to FIG. 20, after receiving a partial match search request for “hiro”, after excluding entries in which “Hikoroku” that includes “hi” and “ro” but is not continuous, “Hiroshi” and “Hiroroku” Only entries with "Chihiro" registered can be search hit candidates. It is apparent that a more convenient directory search method can be provided by a combination of the above embodiments.

As described above, according to the character string search device of the present embodiment, the character string is searched based on the reading of the keyword, so that the usability of the character string search can be improved. A program for realizing the present invention may be stored in a recording medium, and the program may be installed and used from the recording medium using a computer (or a portable terminal), or a network may be stored in the recording medium. The program may be accessed and downloaded to use the program.

[0069]

According to the present invention, since a character string is searched based on the reading of a keyword, the usability of the character string search can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a schematic configuration of a system that provides a directory service according to a first embodiment.

FIG. 2 is a diagram illustrating an example of an information configuration of a reading matching index 107 according to the first embodiment.

FIG. 3 is a diagram illustrating an information configuration example of index definition information 105 according to the first embodiment.

FIG. 4 is a flowchart illustrating a processing procedure of a directory control unit according to the first embodiment.

FIG. 5 is a flowchart illustrating a processing procedure of a directory update control unit 111 according to the first embodiment.

FIG. 6 is a flowchart illustrating a processing procedure of an index update control unit 113 according to the first embodiment.

FIG. 7 is a flowchart illustrating a processing procedure of a pattern matching index update unit 115 according to the first embodiment.

FIG. 8 is a flowchart illustrating a processing procedure of a pattern matching key extraction unit 110 according to the first embodiment.

FIG. 9 is a reading match index updating unit 11 according to the first embodiment;
6 is a flowchart illustrating a processing procedure of No. 6;

FIG. 10 is a flowchart illustrating a processing procedure of a reading matching key extraction unit 109 according to the first embodiment.

FIG. 11 is a diagram illustrating an example of a Roman alphabet notation table according to the first embodiment.

FIG. 12 is a diagram illustrating an example of a Roman character normalization table according to the first embodiment.

FIG. 13 is a diagram illustrating a screen display example at the time of attribute search according to the first embodiment.

FIG. 14 is a directory search control unit 112 according to the first embodiment.
6 is a flowchart showing the processing procedure of FIG.

FIG. 15 is a diagram illustrating a screen display example of an attribute search result according to the first embodiment.

FIG. 16 is an index search control unit 114 according to the first embodiment.
6 is a flowchart showing the processing procedure of FIG.

FIG. 17 is a flowchart illustrating a processing procedure of a pattern matching index search unit 117 according to the first embodiment.

FIG. 18 is a reading match index search unit 1 according to the first embodiment;
It is a flowchart which shows the processing procedure of 18.

FIG. 19 is a diagram illustrating an information configuration example of a reading matching index 107 according to the second embodiment.

FIG. 20 is a diagram illustrating an information configuration example of a reading matching index 107 according to the third embodiment.

FIG. 21 is a diagram showing an example of the information configuration of a directory DB in which a conventional directory service stores each directory entry.

FIG. 22 is a diagram illustrating an information configuration example of a pattern matching index for an English name attribute in a conventional index search method.

[Explanation of symbols]

101: directory server, 102: client,
103: network, 104: directory DB, 1
05 index definition information, 106 pattern matching index, 107 reading matching index, 108
Directory control unit, 109: reading matching key extracting unit, 1
10: pattern matching key extraction unit, 111: directory update control unit, 112: directory search control unit, 113
... Index update control unit, 114 ... Index search control unit, 115 ... Pattern matching index update unit, 1
16: reading matching index updating unit, 117: pattern matching index searching unit, 118: reading matching index searching unit, 119: syllable element extracting unit, 120: syllable element normalizing unit, 121: key element extracting unit, 122: pattern normal , 2001: reading match index table, 200
2 ... read matching key, 2003 ... entry ID list, 1
701: Index definition information, 1801: Roman alphabet notation table, 1901: Roman alphabet normalization table, 19
02 area, 1903 area, 1301 screen, 130
2 ... Area for selecting an attribute name, 1303 ... Area for designating a search pattern character string, 1304 ... Area for selecting a reading match search or a pattern match search, 1305 ... Area for selecting a partial match or perfect match, 1306 ... OK button , 1307
... Cancel button, 1401 screen, 2101 reading matching index table, 2102 reading matching key, 210
3 Entry ID list, 2201 reading matching index table, 2202 reading matching key, 2203 entry ID list, 1501 directory DB configuration example, 16
01: pattern matching index table, 1602: pattern matching key, 1603: entry ID list.

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Toshiharu Nakamura 5030 Totsuka-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture F-term in the Software Division, Hitachi, Ltd. F-term (reference) 5B075 ND03 NK02 NK32 NK54 QP10 QS20

Claims (5)

[Claims] 1. A character string search method for searching a character string, comprising: a step of extracting a syllable element from an input keyword; a step of normalizing the notational fluctuation of the extracted syllable element; And searching for a character string that matches the input keyword reading by referring to a reading matching index indicating a storage location of the character string that contains the character string. 2. The character string search method according to claim 1, wherein a full-match character string search is performed to search for a character string that matches the normalized keyword obtained by connecting the normalized syllable elements. . 3. The character string search method according to claim 1, wherein a character string including a plurality of continuous syllable elements is searched for in the normalized syllable elements. 4. A syllable element extracting unit for extracting a syllable element from an input keyword, and a syllable element normalizing unit for normalizing the notation fluctuation of the extracted syllable element. And a reading matching index search unit that searches for a character string that matches the reading of the input keyword by referring to a reading matching index indicating a storage location of a character string including the normalized syllable element. Character string search device. 5. A syllabic element extracting unit for extracting a syllable element from an input keyword on a computer-readable recording medium storing a program for causing a computer to function as a character string searching device for searching for a character string. The syllable element normalization unit for normalizing the fluctuation of the notation of the extracted syllable element, and the reading matching index indicating the storage location of the character string including the normalized syllable element, refer to the reading of the input keyword. A recording medium having recorded thereon a program for causing a computer to function as a reading match index search unit for searching for a character string.