JP2006164086A - Online knowledge search support system and online knowledge search support method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To support an end user in extracting personal connections placing a high value and certain reliance on each other about target knowledge according to category information classifying entries and acquaintance information both in a Weblog server 1, and acquiring reliable knowledge from the content of Weblogs belonging to the connections. <P>SOLUTION: A knowledge resource collection apparatus 2 collects a Weblog content from the Weblog server 1, and extracts a set of Weblogs in acquaintanceship and having common category information as personal connections familiar with each category. A knowledge resource search support server 3 can search the Weblog content belonging to connections familiar with search target categories for the Weblog content belonging to a search target category including synonyms, and visualize and display search results on an end user terminal 4 by classification based on categorical connections, Weblogs, a time series, mutual comment relations, mutual trackback relations or the like. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an online knowledge search technology based on acquaintance information and category information, and more particularly to a technology that supports online knowledge search by extracting a weblog community that stores knowledge about a specific category on a network.

In the conventional online knowledge search technology, as a method of further narrowing down the knowledge to be searched from general search results based on keywords and page rank,
(1) Narrowing search results based on topic extraction by natural language processing (see, for example, Patent Document 1).

(2) Narrowing search results by collaborative filtering (see, for example, Patent Document 2).
There is a method.

  In the narrowing-down method according to the above (1), contents referring to the knowledge to be obtained are generally extracted by natural language processing including morphological analysis and syntax analysis.

  Further, in the narrowing-down method according to (2), preference information extraction based on analysis of preference information, bookmarks, etc. registered as part of the user profile in a limited number of collaborative filtering participant users, and search for each participant user Based on the history, the search results are narrowed down using the search history of a finite number of participating users who have similar preferences to the searcher.

Also, weblog servers typically track entries that weblog authors register daily, comments that weblog readers add to the entries, and trackbacks indicating that entries related to the entries in different weblogs are registered. In addition to accumulating web log contents configured, the category information in which entries are classified for each topic and metadata such as web log URLs related to acquaintances on the Internet of the web log author are stored in Foaf (for example, Non-Patent Document 1). Acquaintance information described in a format such as In the present invention, online knowledge search support using these is provided.
JP 2001-325272 A JP 2000-331020 A “FOAF Vocabular Specification”, Dan Brickley & Libby Miller, 2004/09/02 update, 2004/10/19 search, Internet URL <http://xmlns.com/foaf/0.1/>

  In the prior art, in the case of (1), there is a problem that the content including the description relating to the desired knowledge can be extracted but the reliability of the description relating to the knowledge in the content cannot be guaranteed.

  Further, in the case of (2), there is a problem that it is difficult to improve the narrowing accuracy by collaborative filtering by a limited number of end users, and it is not easy to increase the number of end users participating in collaborative filtering.

  The present invention has been made in view of the above problems, and the purpose of the present invention is to categorize category information classified as entries, which is commonly used as metadata in a web log in online knowledge search technology, and a format such as Foaf. Using the acquaintance information described in the above, we will mutually evaluate the target knowledge and extract a human network with a certain level of trust, so that we can acquire highly reliable knowledge from the content of the weblog belonging to this human network. An object of the present invention is to provide an online knowledge search support apparatus and method that can support an end user.

  In order to achieve the above object, the present invention described in claim 1 and the like is characterized by the following apparatus and method.

(1) An apparatus that supports an online knowledge search including a knowledge resource collection device and a knowledge search support server,
The knowledge resource collection device includes:
Crawling means for collecting acquaintance information on the Internet, including entries, comments, trackbacks and category information included in general weblog contents, and weblog information of weblog author acquaintances;
Human network information extracting means for extracting an acquaintance network having a common category from acquaintance information included in the weblog content collected by the crawling means to extract an acquaintance network that is familiar with the category as personal network information for each category;
A knowledge resource extracting means for indexing the web log content belonging to the personal network information in a searchable form for each category and storing it as a knowledge resource;
The knowledge search support server is designated a search target category from browsing software such as a web browser of a general end user terminal, searches the knowledge resource for a category including a synonym of the search target category, Weblog entries included in weblog content, weblog authors, time series and category information, comments, trackbacks, and network information for each category are displayed on the end user terminal after visualizing the structure. An online knowledge search support apparatus characterized by comprising knowledge resource search means for enabling.

(2) A method for supporting online knowledge search comprising a knowledge resource collection device and a knowledge search support server,
The knowledge resource collection device includes:
A crawling process for collecting acquaintance information on the Internet, including entries, comments, trackbacks and category information included in general weblog contents, and web log URLs of acquaintances of weblog authors;
A network information extraction process for extracting a network of acquaintances familiar with the category as network information for each category by extracting an acquaintance relationship having a common category from acquaintance information included in the weblog content collected in the crawling process,
A knowledge resource extraction process of indexing the web log content belonging to the network information in a searchable form for each category and accumulating as a knowledge resource,
The knowledge resource search support server is designated a search target category from browsing software such as a web browser of a general end user terminal, searches the knowledge resource for a category including a synonym of the search target category, and a search result Visualization of the structure of the weblogs, weblog authors, entries categorized by time series and category information, comments, trackbacks, and network information for each category on the end user terminal An online knowledge search support method comprising a knowledge resource search process that enables display.

  As described above, in the present invention, the knowledge resource collection device collects web log contents, extracts a set of weblogs having acquaintance relationships and common category information as detailed personal connections to each category, and knowledge The resource search support server searches the web log contents belonging to the search target category including synonyms from the web log contents belonging to the personal network familiar to the search target category, and the human network for each category, each web log, or time series and mutual comments. In order to make it possible to visualize and display search results by classification of relationships, mutual trackback relationships, etc., we usually use only category information and acquaintance information attached as general metadata to weblogs. A human network that is knowledgeable and trusts enough to maintain mutual acquaintance Out, the content of the weblog belonging to the network of contacts is regarded as reliable with respect to the knowledge makes it possible to support the end-user to the knowledge obtained from the weblog content can be obtained easily.

  Further, since the general metadata of the weblog that is currently popularized can be used as it is, the existing weblog content can be used as it is as a knowledge resource.

  As a result, it is possible to solve the conventional problem that the content including the description related to the desired knowledge can be extracted, but the reliability of the description regarding the knowledge in the content cannot be guaranteed.

  In addition, since existing weblog contents can be used as knowledge resources as they are, it is difficult to increase the accuracy of narrowing down by collaborative filtering by a limited number of end users, and it is not easy to increase the number of end users participating in collaborative filtering. Can solve the problem.

  As described above, according to the present invention, the knowledge resource collection device collects web log contents, extracts a set of weblogs having acquaintance relationships and common category information as detailed personal connections to each category, and knowledge The resource search support server searches web log contents belonging to the search target category including synonyms from the web log contents belonging to the personal network familiar with the search target category, and classifies the human network by category, each web log, or time series. In order to visualize search results and display them on end-user terminals, we usually use only category information and acquaintance information attached as general metadata to weblogs. Extract the human network that has enough trust to maintain the web, and the web that belongs to this human network Content grayed is deemed reliable regarding the knowledge makes it possible to support the end-user to knowledge can easily acquire a determined from the weblog content.

  As a result, it is possible to solve the conventional problem that the content including the description related to the desired knowledge can be extracted, but the reliability of the description regarding the knowledge in the content cannot be guaranteed.

  In addition, since existing weblog contents can be used as knowledge resources as they are, it is difficult to increase the accuracy of narrowing down by collaborative filtering by a limited number of end users, and it is not easy to increase the number of end users participating in collaborative filtering. Can solve the problem.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a diagram showing the configuration and processing flow of an online knowledge search support apparatus and method according to an embodiment of the present invention as set forth in claim 1 and the like.

  The online knowledge search support apparatus and method shown in FIG. 1 includes a knowledge resource collection device 2 and a knowledge resource search support server 3 using a web log server 1 as an information source, and the devices are connected by a network. A general weblog server 1 and a general end user terminal 4 are connected via the Internet.

  The knowledge resource collection device 2 includes crawling means for collecting, as cache information, web log contents including acquaintance information Foaf, category information CAT, entry ENT, comment CMT, and trackback TRBK registered in a general web log server 1; Human network information extracting means for extracting personal network information RSS for each category based on acquaintance information Foaf and category information CAT, entry ENT, comment CMT and track back included in the web log of each web log author belonging to personal network information RSS for each category A knowledge resource that generates TRBK as a knowledge resource by extracting TRBK from the weblog content and storing the entry ENT, the comment CMT, and the trackback TRBK in association with the personal network information for each category. And a scan extraction means.

  FIG. 2 is a diagram showing a configuration example of general weblog content. In general, entries are text information and image URLs classified by time series and category information, and correspond to what is written in a diary. A comment is usually text information written by a weblog reader for an entry, and may include an image URL or the like. The trackback is a trackback source entry related to the trackback destination entry, is referred to based on the trackbacking received by the trackback destination web log, and often exists in a web log different from the track back destination entry.

  In general, category information is classification information indicating the category, genre, or topic to which the entry belongs, and is often set by the weblog author or weblog server administrator. In addition, general weblog servers have a hierarchical structure. Often not.

  The acquaintance information is usually set for each web log in the Foaf format, and the web log author's name, nickname, mail address or mail address hash value, web log URL, URL of the Foaf itself, etc. are stored in a predetermined XML. In addition to the description under the <Person> tag, reference information to the Foaf of the acquaintance of the weblog author is described under the <Knows> tag, and further, for example, Dubincore (for example, see Non-Patent Document 2 described later). General metadata description is included.

  Here, what is called personal network information RSS for each category sequentially refers to the acquaintance's weblog described in the <Knows> tag in the Foaf of each weblog, and common category information in the weblog group. This is a result of extracting a set of weblogs having. As a simple example of the implementation of network information for each category, for example, the format of RSS 1.0 (for example, see Non-Patent Document 3 described later) is used, and common category information is described in the <description> tag of the <Channel> tag. Then, the reference URL to each Foaf having the category information in common and having a reference relationship with the <Knows> tag is described under the <item> tag. FIG. 3 shows an example of network information for each category.

  Returning to FIG. 1, what is called a knowledge resource is a database that stores, for example, a cache of web log content. The entry ENT, the comment CMT, and the trackback TRBK constituting the web log content, and each entry ENT The category information CAT to which the entry belongs, the web log to which each entry belongs, the author of the web log, the personal network information RSS for each category, and the like, can be searched using these as keys.

  The knowledge resource search support server 3 resolves the synonym based on the synonym dictionary for the search target category requested from the end user terminal 4, and then indexes the entry ENT and the comment associated with the category. A format in which a set of CMTs and trackbacks TRBK is searched from the knowledge resources extracted by the knowledge resource extraction means, and the search results classified in association with the personal network information RSS can be displayed on the end user terminal 4, for example, a WWW browser Knowledge resource search means for returning in html format that can be displayed in The search target category is, for example, a combination of a plurality of keywords.

  Next, the operation of the online knowledge search support apparatus and method of this embodiment configured as described above will be described with reference to step numbers S1 to S4 shown in FIG.

  First, the knowledge resource collection device 2 starts from a general web log server 1 which is connected to the Internet and is open to public, including web acquaintance information Foaf, category information CAT, entry ENT, comment CMT, and trackback TRBK. The web log contents are similarly collected as the cache 2A for the web logs created by the acquaintances of the web log authors described by the <Knows> tag of each Foaf (step S1).

  Next, in the cache 2A, the knowledge resource collection device 2 applies a human network of the category to a set of web logs including the same category information among a plurality of web logs having a reference relationship by the <Knows> tag of Foaf. The information RSS2B is extracted (step S2), and the weblog content belonging to the category of human network information RSS included in the cache 2A is stored in association with the human network information RSS of each category. (Step S3).

  Next, the knowledge resource search support server 3 that has received a plurality of keywords as the search target category from the end user terminal 4 resolves the synonym for each keyword based on the synonym dictionary and then associates the keyword with the keyword. Search the knowledge resource 2C extracted by the knowledge resource extraction means, and the entry ENT classified by the weblog, weblog author, time series and category information included in the weblog content of the search result, The comment CMT, the trackback, and the personal network information RSS for each category described in the RSS format are visualized in the html format as a tree structure or a graph structure, and are returned to the end user terminal 4 (step S4).

[Non-Patent Document 2]
"Dublin Core Metadata Element Set, Version 1.1: Reference Description", updated 2003/06/02, [Search 10/28/2004], Internet URL <http://dublincore.org/documents/dces/>
[Non-Patent Document 3]
"RDF Site Summary (RSS) 1.0", Gabe Beged-Dov & Dan Brickley et.al., 2001/05/30 update, [Search 10/29/2004], Internet URL <http://web.resource.org /rss/1.0/spec>

The figure which shows the structure of the online knowledge search assistance apparatus and method which concern on one Embodiment of this invention, and the flow of a process. The figure which shows the structural example of a general web log content. The figure which shows the example of the personal network information for every category in one Embodiment of this invention.

Explanation of symbols

1 Weblog Server 2 Knowledge Resource Collection Device 3 Knowledge Resource Search Support Server 4 End User Terminal

Claims (2)

An apparatus for supporting online knowledge search comprising a knowledge resource collection device and a knowledge search support server,
The knowledge resource collection device includes:
Crawling means for collecting acquaintance information on the Internet, including entries, comments, trackbacks and category information included in general weblog contents, and weblog information of weblog author acquaintances;
Human network information extracting means for extracting an acquaintance network having a common category from acquaintance information included in the weblog content collected by the crawling means to extract an acquaintance network that is familiar with the category as personal network information for each category;
A knowledge resource extracting means for indexing the web log content belonging to the personal network information in a searchable form for each category and storing it as a knowledge resource;
The knowledge search support server is designated a search target category from browsing software such as a web browser of a general end user terminal, searches the knowledge resource for a category including a synonym of the search target category, Weblog entries included in weblog content, weblog authors, time series and category information, comments, trackbacks, and network information for each category are displayed on the end user terminal after visualizing the structure. An online knowledge search support device characterized by comprising knowledge resource search means for enabling. A method for supporting online knowledge search comprising a knowledge resource collection device and a knowledge search support server,
The knowledge resource collection device includes:
A crawling process for collecting acquaintance information on the Internet, including entries, comments, trackbacks and category information included in general weblog contents, and web log URLs of acquaintances of weblog authors;
A network information extraction process for extracting a network of acquaintances familiar with the category as network information for each category by extracting an acquaintance relationship having a common category from acquaintance information included in the weblog content collected in the crawling process,
A knowledge resource extraction process of indexing the web log content belonging to the network information in a searchable form for each category and accumulating as a knowledge resource,
The knowledge resource search support server is designated a search target category from browsing software such as a web browser of a general end user terminal, searches the knowledge resource for a category including a synonym of the search target category, and a search result Visualization of the structure of the weblogs, weblog authors, entries categorized by time series and category information, comments, trackbacks, and network information for each category on the end user terminal An online knowledge search support method comprising a knowledge resource search process that enables display.

Images