JP2001265774A - Method and device for retrieving information, recording medium with recorded information retrieval program and hypertext information retrieving system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To decrease a probability for retrieval to be missed by improving accuracy in the calculation of suitability to retrieval conditions inputted by a user in an information retrieval system with a hypertext information system as a target. SOLUTION: An anchor extracting part 1 extracts an anchor expressing a link from a structured text comprising a certain page to the other page. An index preparing part 2 stores even an anchor text on the other page expressing the link to a certain page in a data base 3 as position attribute information in addition to the context of the structured text comprising that page by performing indexing added with a position attribute. While using the position attribute information stored in the data base 3, a retrieval processing part 4 retrieves a page suited to the applied retrieval conditions. Concerning the retrieves result found by the retrieval processing part 4, a suitability calculating part 5 calculates the suitability to the retrieval conditions while considering the position attribute information.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval system for indexing a group of documents to be retrieved in an information retrieval system, and for retrieving documents that meet a retrieval condition input by a user.

[0002]

2. Description of the Related Art An information search apparatus searches for a document that satisfies conditions entered by a user, and presents a list of the documents to the user.
However, some of the search results strongly match the search intention, while others hardly match the search intention. Therefore,
Techniques have been developed to assist users in browsing and using search results by calculating relevance to search conditions and presenting them in order of relevance for each document that is a search result.

[0003] An information retrieval apparatus is based on a certain retrieval model, and includes information such as the number of appearances of a word in a document, the size of the document, the total number of appearances of a word in the entire search target document set, the appearance position, and the attribute of the appearance position. Based on the information, the relevance between the search condition and the document is calculated. Typical search models include a vector space type and a probability type (references:
R. Baeza-Yates and B. Ribeiro-Neto, Modern Informatio
n Retrieval, Addison Wesley, 1999.).

On the other hand, in today's document information systems, documents are often managed in a data structure called hypertext in order to facilitate description and browsing of documents. In hypertext, a document is composed of a page composed of a group of text information and links connecting related pages. Although the realization method differs depending on the hypertext information system, a method is generally used in which a page is represented by structured text and a word in the text is given a role as a link to another page. Hereinafter, a portion of a phrase having a link function in the structured text is referred to as an anchor.

[0005] The purpose of an information retrieval system for hypertext is to present a list of pages satisfying a retrieval condition to a user. Therefore, it is customary to treat one page in the hypertext as one document and perform the same processing as that of a general information search system.

[0006]

As described above, in the conventional hypertext information retrieval system, the information used for calculating the fitness is completed in page units. However, in the hypertext information system, while the number of pages to be searched increases, the information amount of text included in each page often decreases.
Therefore, the conventional method of calculating the degree of fitness cannot calculate the degree of fitness with high reliability.

Further, in the conventional information retrieval system, the page which does not include the phrase specified in the retrieval condition has a relevance of 0, which results in retrieval omission. If there are many pages with a small amount of text, search omissions often occur. In addition, a network-compatible hypertext information system such as WWW has a problem in that pages are created independently by many authors, and terms are not controlled, so that search omissions are likely to occur.

SUMMARY OF THE INVENTION An object of the present invention is to provide an information retrieval system for a hypertext information system in which the accuracy of calculating the degree of relevance to a retrieval condition input by a user is improved and the probability of occurrence of retrieval omission is reduced. Method,
It is an object to provide a device and a recording medium on which an information search program is recorded.

[0009]

An information retrieval apparatus according to the present invention inputs a structured text constituting a page to be searched, interprets the structured text, extracts an anchor, and outputs the anchor. Receiving an address of a reference page of the link to be represented, an anchor extracting means for outputting an anchor text used for the visual representation of the link, a database, a structured text constituting a certain page, and an address of the page, Receiving the reference address and the anchor text, interpreting the anchor text as a part of the structured text, outputting position attribute information necessary for search processing, receiving index creation means for storing in the database, and receiving search conditions Reading the position attribute information corresponding to the phrase included in the search condition from the database, Search processing means for obtaining position attribute information of a page that satisfies the search condition, outputting the position attribute information together with the search condition, adding a degree of conformity to an address of a page that satisfies the search condition, and outputting the result as a search result, A fitness calculating unit that receives the search condition and the position attribute information and calculates the fitness.

According to the present invention, when a phrase included in an anchor text expressing a link of a reference source page is specified as a search condition, not only the reference source page where the anchor text exists but also the reference destination page is included in the search condition. Is determined as a page that satisfies the condition, and the word in the anchor text of the referring page is used for calculating the fitness of the referenced page.

[0011]

Next, embodiments of the present invention will be described with reference to the drawings.

Referring to FIG. 1, an information retrieval apparatus according to one embodiment of the present invention includes an anchor extraction unit 1, an index creation unit 2, a database 3, a search processing unit 4, and a fitness calculation unit 5.

As shown in FIG. 2, the anchor extracting unit 1
The structured text constituting the page to be searched is received from the hypertext information system (step 1).
1) Interpret the structured text (step 12) and extract anchors (step 13). Among the information constituting the extracted anchor, the address of the reference page of the link represented by the anchor and the anchor text used for the visual representation of the link are sent to the index creating unit 2 (step 14).

As shown in FIG. 3, the index creating unit 2 receives the structured text constituting a certain page and the address of the page from the hypertext information system (step 21). Also, from the anchor extraction unit 1, a reference address of the link represented by the anchor and the anchor text are received (step 22), and the anchor text is interpreted as a part of the structured text corresponding to the reference address (step 22). 23), generates position attribute information and passes it to the database 3 (step 24).

As shown in FIG. 4, the search processing unit 4 receives the search condition from the search reception unit (step 31), interprets the search condition (step 32), and responds to the words included in the search condition from the database 3 (step 32). The position attribute information to be searched is read out (step 33), and the position attribute information of the page that satisfies the search condition is obtained by collating with the search condition (step 3).
4), and pass it along with the search condition to the fitness calculation unit 5 (step 35). Then, the relevance calculated by the relevance calculator 5 is input (step 36), the relevance is added to the address of the page satisfying the search condition, and the result is passed to the search receiving unit as a search result (step 37).

As shown in FIG. 5, the relevance calculating unit 5 receives the search condition and the position attribute information from the retrieving processing unit 4 (step 41), and calculates the relevance using the appearance frequency of the word (step 42). ), And a more detailed calculation of the degree of conformity is performed using the position attribute information (step 43). here,
The conformity based on the attribute of the reference source anchor is also evaluated.

FIG. 6 is a block diagram showing a case where the above-described information retrieval apparatus of the present invention is applied to an information retrieval system for WWW which is one of the hypertext information systems.

The WWW information search system 100 includes an anchor extraction unit 1, an index creation unit 2, a database 3, a search processing unit 4, a fitness calculation unit 5 (the above constitutes the information search device of FIG. 1), and a search reception unit 6. , A page collection unit 7, a page storage unit 8, and an anchor storage unit 9.

Prior to the search process, the WWW information search system 100 performs an index creation process required for the search process. The page collection unit 7 accesses a WWW server having a page to be searched in accordance with a means conforming to the HTTP protocol, and sends the HTML information from the WWW information system 200 to the HTML.
The structured text in the format is obtained and stored in the page storage unit 8. The HTML-format text stored in the page storage unit 8 is first passed to the anchor extraction unit 1, which extracts only the anchor part from the HTML-format text according to the rules of HTML, and stores them in the anchor storage unit 9. To be stored. Here, the anchor portion is H
It corresponds to the A element defined in the TML rules,
The HREF attribute of the element is the reference address of the link represented by the anchor, and the visible portion of the element described between the start tag and the end tag of the A element becomes the anchor text (Reference: T. Berners) -Lee and D. Connolly,
Hypertext Markup Language-2.0, Request for Comment
s 1866, Internet Engineering Task Force, 1995.).
The index creating unit 2 receives the structured text in the HTML format constituting one page from the page storage unit 8 and
It interprets according to the L rule, sets the appearance position of the phrase in the text based on the address of the page, adds attributes such as a title, a headline, and a keyword, outputs it as position attribute information, and passes it to the database 3. Also, the index creation unit 2 receives the anchor stored in the anchor storage unit 9 as an input, and sets the appearance position of the phrase in the anchor text based on the address of the reference page of the link represented by the anchor, It is output together with an attribute called a reference source anchor, and passed to the database 3. The database 3 stores the position attribute information output from the index creation unit 2 in a format suitable for search processing.

The search processing in the WWW information search system 100 is started when the user 300 inputs search conditions. The search receiving unit 6 receives the search condition from the user 300 and passes it to the search processing unit 4. The search processing unit 4 obtains position attribute information that satisfies the search condition from the position attribute information stored in the database 3 and passes the position attribute information together with the search condition to the matching degree calculation unit 5. From the position attribute information, the matching degree calculation unit 5 obtains the total number of appearances of the search condition word, the number of appearances of the search condition word on each page satisfying the search condition, the attribute at the appearance position of the search condition word, and the like. The relevance is calculated and passed to the search processing unit 4. At this time, weighting is performed so that the attribute of the reference source anchor contributes more to the degree of matching with the search condition than the standard attribute. The search processing unit 4 obtains addresses of pages satisfying the search conditions, and transfers them to the search reception unit 6 as a search result, in which the addresses are rearranged in the order of the magnitude of the fitness calculated by the fitness calculation unit 5. The search receiving unit 6 passes a part or all of the search results arranged in the order of relevance to the user 300,
The search processing ends.

Referring to FIG. 7, an information retrieval apparatus according to another embodiment of the present invention comprises an input device 51, storage devices 52 and 53, an output device 54, a recording medium 55, and a data processing device 56.

The input device 51 is an input device such as a keyboard for inputting structured text, search conditions and the like.
The storage device 52 corresponds to the database in FIG. The storage device 53 is a hard disk. The output device 54 is an output device, such as a display or a printer, to which a search result is output. The recording medium 55 is a floppy disk, CD-ROM, magneto-optical disk in which an information retrieval program including the anchor extraction unit 1, index creation unit 2, search processing unit 4, and fitness calculation unit 5 in FIG. It is a recording medium such as a disk. The data processing device 56 includes a CPU, various interfaces, etc., reads an information retrieval program from the recording medium 55 into the storage device 53, and executes the program.

[0023]

As described above, according to the present invention,
When a search is performed using search conditions that include the term contained in the anchor text, the page referred to by the link represented by the anchor is also a search result. It is possible, and the probability of occurrence of search omission is reduced. Also. Since the anchor text often expresses the outline of the referenced page simply and accurately, it is possible to calculate a more appropriate degree of matching than when only the text in the page is used. In a network type hypertext information system such as WWW, a page containing useful information is:
Although it tends to be linked from many pages, in the hypertext information search system to which the present invention is applied,
Pages referenced by anchors containing common words from many pages will have high relevance, and consequently pages containing useful information will be displayed at the top of the search results sorted by relevance Therefore, the convenience of the search system user is improved.

For example, the official homepage of Nippon Telegraph and Telephone Corporation is linked from many WWW pages by anchor texts such as "NTT" and "Nippon Telegraph and Telephone". In the WWW information search system of the present invention, "N
When the word “TT” is specified as a search condition, the relevance of such a page is increased, and the search results are displayed in priority when the search results are arranged in the order of relevance, thereby improving user convenience. I do. Also, even if the word “NTITY” is not used at all on its official website, if it is linked with an anchor text containing the word “NTITY”, the search condition will be “NTITY”. Is included in the search result, and the probability of occurrence of the search omission is reduced.

If the information retrieval apparatus uses attribute information such as a title, a headline, and a keyword for calculating the degree of matching,
The present invention can be easily applied by adding a new attribute called a reference source anchor. At this time, the contribution of the attribute of the reference source anchor to the relevance is made variable, so that the relevance calculation using only the text in the page as in the past and the relevance calculation considering the reference source anchor Can be used properly.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an information search device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a process of an anchor extracting unit 1 used in the information search device of FIG.

FIG. 3 is a flowchart showing a process of an index creation unit 2 used in the information search device of FIG. 1;

FIG. 4 is a flowchart showing a process of a search processing unit 4 used in the information search device of FIG. 1;

FIG. 5 is a flowchart showing a process of a fitness calculating unit 5 used in the information search device of FIG. 1;

FIG. 6 is a diagram showing a configuration when the information search device of the present invention is applied to a WWW information search system.

FIG. 7 is a block diagram showing an information search device according to another embodiment of the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Anchor extraction part 2 Index creation part 3 Database 4 Search processing part 5 Fitness calculation part 6 Search reception part 7 Page collection part 8 Page storage part 9 Anchor storage part 11-14, 21-24, 31-37, 41-44
Step 100 WWW information search system 200 WWW information system 300 User 51 Input device 52, 53 Storage device 54 Output device 55 Recording medium 56 Data processing device

Claims (4)

[Claims] 1. An information retrieval method for a hypertext, comprising inputting structured text constituting a page to be searched, interpreting the structured text, extracting an anchor, and extracting the extracted anchor. Receiving an address of a reference page of a link represented by an anchor, an anchor extraction step of outputting an anchor text used for visual representation of the link, receiving a structured text constituting a certain page and an address of the page, and Receiving an address and an anchor text, interpreting the anchor text as a part of the structured text, outputting position attribute information necessary for search processing, and storing an index in a database; receiving search conditions; The position attribute information corresponding to the phrase included in the condition is read from the database, A search processing step of obtaining position attribute information of a page that satisfies the search condition by matching the search condition, and outputting the position attribute information together with the search condition; a fitness calculation step of receiving the search condition and the location attribute information and calculating the fitness. An information search method including a search result output step of adding the relevance to an address of a page satisfying the search condition and outputting the result as a search result. 2. An information search apparatus for a hypertext, comprising inputting structured text constituting a page to be searched, interpreting the structured text, extracting an anchor, and extracting the extracted anchor. Receiving the address of the page referred to by the link represented, the anchor extracting means for outputting the anchor text used for the visual representation of the link, a database, the structured text constituting a certain page, and the address of the page, Further, receiving the reference destination address and the anchor text, interpreting the anchor text as a part of the structured text, outputting position attribute information necessary for search processing, indexing means for storing in the database, Receiving and reading the position attribute information corresponding to the phrase included in the search condition from the database. Then, by matching with the search condition, the position attribute information of the page satisfying the search condition is obtained, output together with the search condition, the relevance is added to the address of the page satisfying the search condition, and the search is output as the search result. An information retrieval apparatus comprising: a processing unit; and a fitness calculating unit that receives the search condition and the position attribute information and calculates the fitness. 3. A recording medium on which an information search program for hypertext is recorded, wherein a structured text constituting a page to be searched is input, the structured text is interpreted, and an anchor is extracted. , The address of the page referred to by the link represented by the extracted anchor, the anchor extraction process of outputting the anchor text used for the visual representation of the link, the structured text constituting a certain page, and the address of the page Receiving the reference address and the anchor text, interpreting the anchor text as a part of the structured text, outputting position attribute information necessary for search processing, and storing the index information in a database; And from the database, position attribute information corresponding to a phrase included in the search condition. A search result that obtains position attribute information of a page that satisfies the search condition by reading and matching with the search condition, outputs the position attribute information together with the search condition, adds a degree of conformity to an address of the page that satisfies the search condition, and outputs the result as a search result. A recording medium recording an information retrieval program for causing a computer to execute an output process and a fitness calculation process for receiving the search condition and the position attribute information and calculating the fitness. 4. A page storage unit, a page collection unit that acquires a structured text from a hypertext information system and stores the structured text in the page storage unit, an anchor storage unit, and a structured text that constitutes a page to be searched. Is input from the page storage means, the structured text is interpreted, an anchor is extracted, and the address of the reference page of the link represented by the extracted anchor and the anchor text used for the visual representation of the link are described above. An anchor extracting means to be stored in an anchor storing means, a database, a structured text constituting a certain page, and an address of the page are read out from the page storing means, and the reference destination address and the anchor text are read from the anchor storing means. Read the anchor text as part of the structured text Index generating means for outputting position attribute information necessary for search processing and storing the information in the database; search receiving means for receiving search conditions from a user and presenting search results to the user; and receiving the search conditions. Receiving a search condition from the means, reading out position attribute information corresponding to a phrase included in the search condition from the database, and comparing the search condition with the search condition to obtain position attribute information of a page satisfying the search condition;
A search processing unit that outputs together with the search condition, adds a relevance to an address of a page that satisfies the search condition, and outputs the result to the search reception unit as a search result; A hypertext information retrieval system having a fitness calculating means for calculating.