JP2002342345A - Keyword classifying method and terminal equipment - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a terminal equipment, with which the storage areas in a dictionary for classification are reduced and the maintenance of the dictionary for classification is not required. SOLUTION: The terminal equipment 30 sends a keyword to a search server 10 (search engine) and requests the search of a Web site corresponding to the keyword. An analytic part 33 receives the list of Web sites corresponding to the keyword from the search server 10, performs the analysis of required data and extracts category candidates. On the basis of predetermined rules, a judging part 34 selects a category out of the category candidates extracted by the analytic part 33.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a keyword classification method and a terminal device, and more particularly, to a WWW (World).
The present invention relates to a keyword classification method for obtaining a category for a keyword using a search site on Wide Web, and a terminal device using the method.

[0002]

2. Description of the Related Art Conventionally, various types of data classification systems exist. For example, JP-A-6-34875
The document classification system disclosed in Japanese Patent Publication No. 5 detects words in a sample document of each category and registers words appearing in only one category in the classification dictionary as keywords of that category. When classifying a document, the classification dictionary created in this way is used to detect the number of matches between the words included in the classification target document and the keywords registered in the classification dictionary. Many categories are set as the classification destinations of the classification target documents. This eliminates the need for manually defining classification information for classifying documents.

In the merchandise sales information analysis system disclosed in Japanese Patent Application Laid-Open No. 2000-11044, registration of a classification code of a merchandise for analysis is processed, and information trend analysis of sales performance of the merchandise is performed. Next, the classification code of the product for analysis is registered from the analysis condition set value specified in the information trend analysis process of sales performance, and the data in the product classification master database for analysis is registered based on the regularity detected from the product sales performance information. Construction is performed.

[0004]

However, Japanese Patent Laid-Open No. 6-3 / 1994
In the document classification system described in Japanese Patent No. 48755, when automatically creating a classification dictionary for classifying data, it is necessary for a user to create document classifications in advance and store sample documents in a database for each classification. . For this reason, human labor is still required. Further, since a classification dictionary is stored in the system, an extra storage area is required.

On the other hand, in the merchandise sales information analysis system described in JP-A-2000-11044, an analysis merchandise classification master database is constructed on a server. Therefore, the system configuration of the server becomes large. In addition, the analysis product classification master database must be updated every time an analysis request is made from the workstation, which increases the load on the server.

The present invention has been made in view of such a situation, and an object of the present invention is to provide a keyword classification method and a terminal device in which the storage area of a classification dictionary is reduced.

It is another object of the present invention to provide a keyword classification method and a terminal device which do not require maintenance of a classification dictionary.

[0008]

A keyword classification method according to an aspect of the present invention is used in a terminal device connected to a network. The keyword classification method includes a step of designating a search condition, a step of referring to a search engine that provides, as a home page, a hierarchical category to which the search condition belongs and a link destination of a home page that matches the search condition, and forming the home page from the search engine Obtaining text data, analyzing the obtained text data to extract category candidates, and determining a category that matches a search condition from the extracted category candidates based on a predetermined rule. Including.

A category is obtained by using a search engine. Therefore, it is not necessary to have a database unique to the system, and the system can be simplified.
In addition, since a storage area for the classification dictionary is not required, the storage area can be reduced. Further, categorization in search engines is regularly maintained. Therefore, it is not necessary for the user to maintain the classification dictionary, and the latest classification information can always be obtained.

[0010] Preferably, the step of determining a category includes a step of determining a category having the highest appearance frequency in the text data among the category candidates.

[0011] A terminal device according to another aspect of the present invention is connected to a network. The terminal device specifies a search condition, refers to a search engine that provides, as a homepage, a hierarchical category to which the search condition belongs and a link destination of a homepage that matches the search condition, and outputs text data constituting the homepage from the search engine. A search unit for acquiring, and an analysis unit connected to the search unit for analyzing text data to extract a category candidate; and an extracted category candidate connected to the analysis unit based on a predetermined rule. And a judging means for judging a category that matches the search condition from the above.

A category is obtained by using a search engine. Therefore, it is not necessary to have a database unique to the system, and the system can be simplified.
In addition, since a storage area for the classification dictionary is not required, the storage area can be reduced. Further, categorization in search engines is regularly maintained. Therefore, it is not necessary for the user to maintain the classification dictionary, and the latest classification information can always be obtained.

Preferably, the judging means includes means for judging a category having the highest appearance frequency in the text data among the category candidates.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Referring to FIG. 1, a search system according to an embodiment of the present invention is connected to a terminal device 30 used by a user and to the terminal device 30 via the Internet 20. HTML (Hyper T
ext Markup Language) data.

A WWW is constructed on the Internet 20. In the WWW, an HTTP (Hyper)
Data transfer is performed by a protocol called “Text Transfer Protocol”. Also,
By describing a Web page in a description language represented by HTML, information can be easily searched and displayed.

The search server 10 includes a WWW server 11 and
A database 12 storing information published on the Internet, a WWW server 11 and a We connected to the database 12 and searching the database 12 based on keywords received from the WWW server 11;
b site search unit 13, connected to the Web site search unit 13 and the WWW server 11,
a web page generation unit 14 that generates a list of websites and sends the list to the terminal device 30 via the WWW server 11. Such a search server 10 is generally called a search engine.

Search engines are generally classified into a full-text search type and a directory type. In the present invention, a directory-type search engine that performs category classification by directory is used. Its famous one is Y
AHOO! (R) and LYCOS (R).

[0018] These directory-type search engines are provided by the creator of the Web site or the administrator of the search engine.
By registering the website, a database is constructed. Since the specific person registers in this way, the number of Web sites registered in the database is small, but the reliability is high. Further, since the database is classified, it is unlikely that an unrelated Web site will be searched. For these reasons, the categories included in the list of Web sites provided by these search engines are handled as data classification dictionaries.

The terminal device 30 is a personal computer or a portable terminal, is connected to the Internet 20, accesses various servers using a WWW browser, and receives information and services. Here, the terminal device 30 receives a search service from the search server 10.

The terminal device 30 is connected to the communication control unit 31 via the Internet 20 to control the transmission and reception of information with the search server 10, and is connected to the communication control unit 31 and receives the Web received via the communication control unit 31. An analysis unit 33 for analyzing a list of sites and extracting category candidates; and an analysis unit 3
3 and a determination unit 34 for selecting a category from the category candidates extracted by the analysis unit 33 based on a predetermined rule, and a classification target data such as purchased product data and document data. Data storage unit 35 to be stored
A search unit 32 connected to the communication control unit 31 and transmitting the search condition to the search server 10 via the communication control unit 31
Connected to the output unit 37 for outputting the result, the search unit 32, the determination unit 34, the data storage unit 35, and the output unit 37,
A processing unit 36 for classifying data stored in the data storage unit 35 based on a user's instruction.

In the following description, the data storage unit 35
It is assumed that a plurality of homepages collected from many Web servers (not shown) connected to the Internet 20 are stored together with titles. Based on the title given to each homepage, the homepage is classified into categories using a search engine.

[0034] The processing unit 36, which has received an instruction from the user, reads the title given to each home page from the data storage unit 35 and outputs it to the search unit 32. The search unit 32
Through the communication control unit 31, the title is used as a search condition,
The request is sent to the search server 10, and a search for the corresponding Web site is requested.

When the search result is sent from the search server 10, the analysis unit 33 sends the search result to the Web server via the communication control unit 31.
Receives a list of b sites (Web pages) and performs analysis. The Web page is text data that constitutes a home page. The analysis unit 33 analyzes the text data, determines a category description portion, and extracts a character string that is a category candidate. The determination unit 34 selects a category from the category candidates extracted by the analysis unit 33 based on a predetermined rule, and sends the category to the processing unit 36. The processing unit 36 assigns the selected category to the home page having the title read from the data storage unit 35 and writes the selected category to the data storage unit 35.

Further, the processing unit 36 analyzes the tendency of collecting home pages from the results classified in this way, and outputs the results to the output unit 37. Output unit 3
Reference numeral 7 processes the result received from the processing unit 36 into a graph or the like and displays it.

Referring to FIG. 2, the keyword classification processing will be described. Here, the classification of purchased products will be described as an example.

A user using the terminal device 30 sets a product name as a keyword, for example, "MD", and requests a search engine to search a Web site (step S).
11). Upon receiving the search request, the Web site search unit 13 of the search engine searches for a Web site that matches the keyword “MD” (Step S12). The terminal device 30
From the search server 10, a W-list including a list of titles and category information of Web sites that hit the keyword "MD"
An eb page is obtained (step S13).

FIG. 3 shows the W obtained from the search server 10.
It is a conceptual diagram of an web page, and is an example of the search result of the web site with respect to the keyword "MD". Search server 1
The data obtained from 0 is actually text data describing the configuration of a Web page in a description language such as HTML, but in FIG. 3, it is represented by a display image on a WWW browser.

The analysis section 33 analyzes the text data (step S14). That is, first, the Japanese character code is converted in order to unify the Japanese character code in the text data. For example, Shift JIS (Japa
EUC (Extended UNI) used in UNIX (R) from nese Industrial Standards) or JIS code etc.
XCode). Next, the analysis unit 33 extracts the category description portion 101 from the text data and divides it into elements. FIG. 4 is a diagram in which the category description portion 101 is divided for each element. Category description part 1 divided
01 includes a category depth 201 and extracted elements, that is, category candidates 202.

The category description portion 101 is delimited by a delimiter such as "/" or ">", and the category candidate 20
2 is extracted. Each time a category candidate 202 is extracted, the depth 201 of the hierarchy increases.

The determining unit 34 determines a category from the category candidates 202 based on a predetermined rule (step S15). As a rule for determining a category, there are various methods such as determining an element in a layer immediately above the element that matches the keyword as a category, or performing statistical processing and determining the category from the positional relationship. Can be Here, the category having the highest number of appearances in the text data is determined as the category corresponding to the keyword, excluding the category element of the lowest layer. For example, in FIG. 4, the lowest layer “M
D ”and“ recording medium ”,“ AV ”
It is determined that “device” is a category for the keyword “MD”. The processing unit 36 outputs “AV device” as a result of the determination to the output unit 37, and ends this processing (step S16).

In this embodiment, one category is automatically determined from a plurality of category candidates, but one keyword is set so as to be included in the plurality of categories. Is also good.

Instead of automatically determining a category, a user may select a category from a plurality of categories.

As described above, according to the present embodiment, the terminal device 30 uses the search engine (the search server 10) instead of the conventional classification dictionary.
Therefore, it is not necessary to have a database unique to the system, and the system can be simplified.

Further, a storage area for the classification dictionary in the terminal device 30 is not required. Further, categorization in search engines is regularly maintained. For this reason, the terminal device 30 does not need to maintain the classification dictionary, and can always acquire the latest classification information.

The embodiments disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

[0036]

According to the present invention, the storage area of the classification dictionary can be reduced. Further, the maintenance of the classification dictionary is unnecessary, and the latest classification information can be always obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a search system according to an embodiment of the present invention.

FIG. 2 is a flowchart of a keyword classification process.

FIG. 3 is a conceptual diagram of a Web page acquired from a search server 10.

FIG. 4 is a diagram showing an example of an extracted category candidate.

[Explanation of symbols]

10 search server, 11 server, 12 database, 13 site search section, 14 page generation section, 20
Internet, 30 terminal devices, 31 communication control unit,
32 search part, 33 analysis part, 34 judgment part, 35 data storage part, 36 processing part, 37 output part, 101 category description part, 202 category candidate.

Claims (4)

[Claims] 1. A keyword classification method for a terminal device connected to a network, wherein a search condition is specified, and a hierarchical category to which the search condition belongs and a link destination of a homepage that matches the search condition are set as homepages. Referring to a search engine to be provided; acquiring text data constituting a homepage from the search engine; analyzing the acquired text data to extract a category candidate; based on a predetermined rule And determining a category that matches the search condition from the extracted category candidates. 2. The keyword classification method according to claim 1, wherein said step of determining a category includes a step of determining a category having the highest appearance frequency in said text data among said category candidates. 3. A terminal device connected to a network, wherein a search engine is provided for designating search conditions and providing, as a home page, a hierarchical category to which the search conditions belong and a link destination of a home page matching the search conditions. A search unit for acquiring text data constituting a home page from the search engine, and an analysis unit connected to the search unit for analyzing the text data and extracting a category candidate. A terminal connected to the analyzing unit and including a determining unit configured to determine a category that matches a search condition from the extracted category candidates based on a predetermined rule. 4. The terminal device according to claim 3, wherein said determining means includes means for determining a category having the highest frequency of appearance in said text data among said category candidates.