KR100434902B1 - Knowledge base custom made information offer system and service method thereof - Google Patents

Abstract

A knowledge-based customized information retrieval system and a service method thereof are disclosed. The knowledge-based customized information retrieval system according to the present invention includes a personalized information search server providing a search service and at least one or more customer terminals using a search service provided by the personalized information search server, and the personalized information search server searches from a customer terminal. The requested search term is received or periodically, accesses a number of websites that provide information, searches for documents containing the search term, and uses the knowledge-based method for the searched documents to identify the web page structure of the websites. An information collecting unit to construct a document, a search information database storing documents searched by the information collecting unit, and words related to a search word as attribute information, and a garbage document using a knowledge-based method for the search word, the attribute information, and the searched document. Then remove the knowledge base Priority of the document is determined by the differentiated weights of the attribute words set through the service, and the documents arranged in the order of priority by the request of the customer terminal are provided to the customer terminal in real time according to the determined priority. Characterized by providing a personalized information providing unit, it is possible to provide a personalized information more suited to the needs of customers by fastening the real-time search through the information ontology DB, weighting the searched documents in a knowledge-based manner.

Description

Knowledge base custom made information offer system and service method

The present invention relates to a search system for searching and providing information requested from a customer on the Internet. In particular, the present invention searches for information desired by a customer in real time and extracts only high-quality information according to various conditions. The present invention relates to a knowledge-based customized information retrieval system and a service method thereof capable of searching for more accurate information and providing the information as customized information.

A conventional information retrieval system accepts a simple query of a customer and outputs a common result extracted by a common retrieval system to a user. In this case, since the general search system is only a search that divides the area of the query word received from the customer, it is difficult for the customers who receive it to obtain customized information differentiated according to their tastes and characteristics. In addition, in the related art, since the information desired by the customer is searched only for a query input by the customer, the search range is widened, and thus, the accuracy and reliability of the searched information are deteriorated due to a large difference between the information desired by the customer and the extraction result.

In addition, since the conventional information retrieval system performs a task depending on the retrieval system used by the sites providing the information when searching the query received from the customer, the accuracy of the retrieved information is inferior and it is difficult to provide real-time information. However, when a search for information is generated immediately, such as a newspaper article, or after a long time, it is important to search for information in real time.

In addition, the conventional personalized information retrieval system searches for the information in all areas of the site regardless of the specific location of the information to be extracted by proceeding without sorting the information providing sites when searching for information desired by the customer As a result, the speed of information extraction and the efficiency of work have decreased.

Therefore, there is a need for a retrieval system for searching / providing information with high reliability and accuracy in real time.

The technical problem to be achieved by the present invention is to provide a knowledge-based customized information retrieval system that can search and provide more accurate and faster information requested by the customer through the application of knowledge-based weight.

Another object of the present invention is to provide an information retrieval method performed in the knowledge-based customized information retrieval system.

1 is a system diagram showing the overall configuration of a knowledge-based customized information retrieval system according to the present invention.

FIG. 2 is a block diagram illustrating each part of the personalized information search server 14 shown in FIG. 1 in more detail.

3 is a diagram illustrating an example of a screen used when a customer searches for information in the system illustrated in FIG. 1.

4 is a diagram illustrating an example of a screen showing a search result of the personalized information search server 14.

FIG. 5 is a flow diagram illustrating one embodiment of a method for a personalized information search server 14 to retrieve information requested from a customer terminal 10.

FIG. 6 is a flowchart illustrating the knowledge-based weighting step (step 135) in FIG.

In order to achieve the above object, the knowledge-based personalized information search system according to the present invention is connected to a personalized information search server through a personalized information search server and communication network providing a search service, at least using a search service provided by the personalized information search server Equipped with one or more customer terminals, the personalized information search server receives a search request requested from the customer terminal or periodically, access a plurality of websites providing information to search for documents containing the search terms, and retrieved documents An information collecting unit for constructing a web page structure of websites as an information ontology using a knowledge-based method, a search information database storing documents retrieved by the information collecting unit, and words related to a search word as attribute information, Search terms, attribution information, and After removing the garbage document using the knowledge-based method, the document is prioritized according to the differentiated weights of the attribute words set through the knowledge-based method, and provided to the client terminal in real time according to the determined priority. It includes a personalized information providing unit for providing to the customer terminal the documents arranged in priority order by the request of the customer terminal.

In order to achieve the above another problem, in a computer network system in which a personalized information search server providing an information search service and at least one customer terminal provided to each customer are connected through a communication network, knowledge-based personalized information through a personalized information search server The service method may refer to the information ontology if the information ontology is constructed in the steps (a) of receiving a search request requested from the customer terminal and determining whether the information ontology for the search term has been constructed. (C) step of accessing the information about the search term from the information providing websites in real time and performing a real-time search, and storing the searched document in the search information database. Remove garbage documents according to the importance of attribute information In addition, the information ontology is constructed in steps (d) and (b) of determining the priority of the searched document by weighting each document through a knowledge-based method, and providing the searched document to the customer terminal according to the priority. If not, the information structure of the document retrieved in steps (e) and (e) accessing the URLs of the information providing websites in real time to search the entire information providing websites in real time, and storing the retrieved document in a search information database. (F) step of constructing the information ontology for the search word by analyzing the.

In order to achieve the above another object, in a computer network system linked to a specific website, a search server for searching and providing information on the linked site, and at least one customer terminal provided to each customer through a communication network, the personalized information In the search method, in step (a) of constructing an information ontology for a linked site, a document search is periodically performed by directly accessing a location where information on a search term is registered in information providing websites by referring to the information ontology and searching for a document. (B) storing the data in the search information database, removing garbage documents according to the importance of attribute information assigned through the knowledge base method from the documents stored in the search information database, and assigning weights to each document through the knowledge base method. (C) and determining the priority of the retrieved documents At the request of the service information from the terminal, made of a (d) providing to the client terminal to extract the stored order to the first document in step (c).

Hereinafter, a knowledge-based customized information retrieval system and a service method thereof according to the present invention will be described with reference to the accompanying drawings.

1 is a system diagram showing the overall configuration of a knowledge-based customized information retrieval system according to the present invention. The personal information search server 14 is a computer provided in a company that hosts a personalized information providing system, and the customer terminal 10 is a computer of a person who receives an information search service using a system operated by the personal information search server 14. . The customer terminal 10 is connected to the customized information retrieval server 14 through a communication network 12 such as the Internet.

The customized information search server 14 is provided with a program and a database for searching for information requested by a customer. The customer terminals 10 have Internet browsers (eg, Netscape, Internet Explorer) capable of displaying web content in the form of Hyper Text Markup Language (HTML). An internet browser allows each computer to access and display the contents of the information retrieval HTML template in the customized information retrieval server 14. The HTML template of the custom information retrieval server 14 has a main web (WWW) page to display to the user.

Referring to FIG. 1, in order to use a service provided by a personalized information providing system, the customer terminal 10 first registers his or her personal information, sets an ID and a password, and registers as a member. When the registered customer requests a search for information to the personalized information search server 14 using the customer terminal 10, the information collecting unit 14a searches for and collects information requested by the customer using a predetermined search driver. do. Here, the information requested to be searched by the customer is for all information such as person, institution, company, school, topic. In addition, the personalized information search server 14 targets all information providing websites provided through the Internet such as internet newspapers, internet magazines, internet broadcasting, and articles, and searches URLs of information providing websites that are searched. Locator is stored as a database.

The custom information search server 14 pre-builds the attribute information that can more specifically define the search terms that can be requested from the customer in the knowledge base. When the search request requested through the customer terminal 10 is received through the communication network 12, the personalized information search server 14 searches a plurality of information providing websites to search for documents including the search word, and the search word. Using only the attribute information on, extracts only those documents that are useful to the customer. In addition, the customized information search server 14 assigns a weight to each of the extracted documents according to the attribute information. In this case, the higher the weight, the more accurate the customized information to the customer's needs. In addition, the attribute information for the search word may be built by the customized information search server 14 through repetitive search and may be directly input by an expert. Specifically, the customized information search server 14 includes an information collecting unit 14a, a search information database DB 14b, and a knowledge-based customized information providing unit 14c.

When the search request request from the customer computer 10 is received through the communication network 12, the information collecting unit 14a searches a plurality of information providing websites, searches for documents including the search word, and searches the searched documents for search information DB. Store in 14b.

The personalized information providing unit 14c is constructing attribute information that can specifically define a search query requested as a knowledge-based DB, and among the documents retrieved and stored in the search-information DB 14b using the knowledge-based DB Only documents that are useful to you are extracted and provided. As described above, the attribute information is information for more specifically defining the search word. If the search word is a specific person, the attribute information is stored along with the person's name, occupation, English name, Chinese character name, and nickname. In addition, if the search term is a specific institution name, along with the institution name, the department name, nickname, English name, Chinese character name, etc. are stored together as attribute information.If the search term is a company name, the department name, nickname, English name, Chinese character name, etc. If the search word is a school name, together with the school name, region, nickname, English name, Chinese character name, etc. are stored together as attribute information. In the case of a topic, important keywords that can specifically define the topic are stored together as attribute information. As such, various attribute information that specifically defines a search word is stored together in the knowledge base, thereby minimizing the retrieval of useless information, that is, garbage information. Can provide. For example, when searching for information on a specific person, the same name person can be easily distinguished by attribute information such as affiliation, occupation, English name, Chinese name, nickname, and the like.

Subsequently, the knowledge-based personalized information providing unit 14c removes the garbage document by using the attribute information and extracts only documents useful to the customer. Then, using the attribute information and the location information of the document, and weight each of the documents useful to the customer. Here, the location information of the extracted document is information on the location of the document extracted from the website providing the extracted document. For example, if a document is extracted from an Internet newspaper, the location of the extracted document may be information reflecting its importance, and accordingly, the weight reflecting the importance may vary according to the location of the document. In addition, the weight of the extracted document may be determined according to how many words representing the attribute information of the search word and the search word stored in the knowledge base DB are related to each other and appear. This weight is a measure of the degree of fit to the customer's needs, the higher the weight is useful information to the customer. In addition, weights may be assigned to each attribute information. That is, even if the word corresponding to the attribute information appears in the extracted document the same number of times, the weight of the extracted document may vary according to the weight of the attribute information. In addition, when words corresponding to high-weighted attribute information are grouped together in the extracted document, the weight is also increased. In this way, the extracted document is analyzed according to the location and attribute information of the document, and the priority of the extracted document is determined.

Meanwhile, the information collecting unit 14a may build an information ontology DB using data stored in the search information DB 14b. Here, the information ontology represents the positional information on the website of the document retrieved by the information collecting unit 14a. For example, in order to search the search term requested by the customer in real time, the information collecting unit 14a accesses the information providing website in real time using the search target URL, and then searches for each section. In addition, the information collecting unit 14a searches for the desired information by searching the information for each depth in each section. Through this procedure, documents including a search word are searched, and the searched documents are stored in the search information DB 14b together with the location information of the documents. Here, the location information may include the website name (URL name, for example, Dong-A Ilbo, Hankyoreh Newspaper, Chosun Ilbo ...), section name (eg, entertainment, sports, entertainment ...), and depth depth). In this case, the depth, for example, in the case of newspapers, is a unit for classifying articles in a vertical structure. The lower the depth, the greater the weight of the article, and the higher the depth, the weight may be classified into the smaller article.

The information collecting unit 14a extracts only the location information of the retrieved article to build an information ontology DB. As such, when the information ontology DB is constructed, when a search for the same search word is requested again in the future, the time required for the information search can be shortened by directly accessing the corresponding location by searching the corresponding document by referring to the information ontology DB. If the information ontology is not established, the information collecting unit 14a takes a lot of searching time because the information collecting unit 14a needs to search all the articles for each section in each section by searching for the corresponding URL.

On the other hand, it has been described above that the personalized information search server 14 searches for information in real time when the information search for the search word is requested from the customer terminal 10 and provides the information to the customer terminal 10. However, the personalized information search server 14 periodically searches for information on a specific search term and stores the information in an internal database. When a request for providing information about a specific search term is requested from the customer terminal 10, the personal information search server 14 refers to the internal database. Data may be extracted to a search term and provided to a customer terminal. For example, when the personalized information search server 14 is linked to the homepage of a specific celebrity, the personalized information search server 14 periodically searches for and stores information about the specific celebrity, and when a request is made to provide information from a customer. Extract and save the contents.

FIG. 2 is a block diagram illustrating each part of the personalized information search server 14 shown in FIG. 1 in more detail. Referring to FIG. 2, the information collecting unit 14a includes a URL list DB 20, a search driver 30, an information ontology DB 40, and an information ontology constructing unit 50. In addition, the knowledge-based custom information providing unit 14c is configured to include a knowledge base DB 60, detailed information DB 70, customer DB 80, knowledge base weighting unit 90.

Referring to FIG. 2, URL names of a plurality of information providing websites are stored in the URL list DB 20. As described above, the information providing website includes all Internet newspapers, Internet magazines, Internet broadcasting, and various documents provided on the Internet, and URL names for them are stored in the URL list DB 20.

The information ontology DB 40 is a database that stores the location of the information requested by the customer. The construction of the information ontology DB 40 analyzes the web page structure of the retrieved information, that is, the URL name, section name, and depth, using the search driver 30, and the analyzed result as ontology information to the information ontology DB 40. Can be built by storing. In addition, experts in the field can update the information ontology. For example, the sports expert may update the information ontology DB 40 by obtaining information on which section of the site the latest information is located through the experts having information on the sports or sports athletes.

The search driver 30 searches the information providing site whose address is the URL stored in the URL list DB 20 and searches for information requested by the customer. Specifically, the search driver 30 includes a meta search driver 32 and a real time search driver 34. Here, the meta search driver 32 searches for information to be searched using a search engine operated by each of the information providing websites listed in the URL list DB 20. As described above, when information is searched using the meta search driver 32, it is used to search for information registered before the search time. The real-time search driver 34 searches in real time the respective information providing websites listed in the URL list DB 20 to search for the latest information. At this time, the real-time search driver 34 refers to the information ontology DB 40 before searching each information providing site, and searches whether the ontology for the information to be searched is constructed. If the ontology of the information to be searched is established, it is possible to directly access the corresponding location, thereby reducing the time required for information search. However, if the information ontology is not established, the search driver 30 needs to search all the articles by the depth of each section of each URL, which takes a lot of searching time.

As such, the search driver 30 may search the information registered in each site and the latest information at the same time before the search time by using the meta search driver 32 and the real time search driver 34, and search the search results DB Store in 14b.

The ontology construction unit 50 extracts only the information indicating the location of the information by using the information stored in the search information DB 14b, constructs the information ontology, stores the result in the information ontology DB 40, and stores the information ontology DB. Update 40. Specifically, the ontology construction unit 50 is configured to include a document structure preprocessor 52 and an ontology generator 54.

The document structure preprocessor 52 extracts only the location information of each retrieved document. The ontology generator 54 generates a document structure from the upper URL to the lower URL using the location information extracted from each document and stores the generated document structure in the information ontology DB 40.

Subsequently, the knowledge base DB 60 of the knowledge base personalized information providing unit 14c stores the search word requested by the customer and the attribute information of the search word. As described above, each of the attribute information may have a weight.

The detailed information DB 70 stores hint information representing the attribute information stored in the knowledge base DB 60 in more detail. For example, first album, second album, concert, album, and the like are hint information about the singer. In other words, if a word such as a first album, a second album, a concert, an album, etc. is detected in the searched document, it may be known that the article is about the singer even if the word singer is not detected in the document.

The customer DB 80 stores information about the customer, that is, a customer ID and a password for each customer.

The knowledge-based weighting unit 90 extracts only the useful documents by removing garbage documents using the attribute information of the retrieved documents stored in the search information DB 14b, and the extracted useful document uses the attribute information and the location information of the document. To give a weight. The knowledge-based weighting unit 90 provides the weighted document to the customer terminal 10 through the communication network 12 as the customized information. In detail, the knowledge-based weighting unit 90 includes a document editing unit 92, a weighting unit 94, and a knowledge-based document DB 96.

The document editing unit 92 refers to the search information DB 14b to extract a document including a search word input by a customer, and removes stopwords such as an investigation from the extracted document. The document is edited using the extracted attribute information by applying the document location and knowledge-based method. Specifically, the document editing unit 92 classifies the extracted document by attribute (alias, region, keyword, hint ...) of the search target, and extracts the document position.

The weighting unit 94 assigns a weight to each document by using the result and position information analyzed according to the attribute information in the knowledge-based document editing unit 92. For example, the priority of the extracted attributes may be determined according to the number of times that the extracted attribute words are mentioned through the knowledge-based method, the extraction relationship among the respective attribute words, the importance of the attribute to which the extracted word belongs, the URL and the depth. Can be. That is, the more the number of attribute words is mentioned, the more the number of related attribute words are closely connected, the higher the importance of the displayed attribute words, and the lower the depth is, the more important the document is recognized and the priority is increased. In addition, the weighting unit 94 determines the priority of the documents according to the weights assigned to the respective documents, sorts the documents according to the priorities, and stores the documents in the document DB 96. Then, as described above, if provided to the customer terminal 10 in real time or if a request from the customer terminal 10, the document is extracted with reference to the document DB and provided to the customer terminal 10.

3 is a diagram illustrating an example of a screen used when a customer searches for information in the system illustrated in FIG. 1. Referring to FIG. 3, when the customer requests information retrieval, the personalized information search server 14 sends the customer name 10 to the customer terminal 10 as shown in FIG. 3. It provides a screen where you can enter "Own", "Keyword", "A pseudonym", "English name" and "Chinese name". The customer inputs "Darkers" in the field "Baseball" and "Belonging" in the field "Park Chan-ho" and attribute information "Occupation" field. In the "Keyword" field, "LA", "Major League", "Striking", "Save", "Pitcher", and other keywords that can represent "Park Chan-ho", and "A pseudonym" in the "Corian Express" After inputting the data in the "English name" and "Chinese name", respectively, and click the "OK" key to transmit the input information to the personalized information search server (14).

4 is a diagram illustrating an example of a screen showing a search result of the personalized information search server 14. The personalized information search server 14 searches for information-providing websites, extracts all articles containing “Park Chan Ho”, and generates weights using the location of the extracted articles and the number of times the attribute information is included in the article contents. 4, the title, weight (W), information source, and date of the article are displayed. In FIG. 4, the personalized information search server 14 sorts the searched articles by date, but may be sorted by weight or by information providing sites. Then, when the customer clicks the desired article among the articles displayed on the customer terminal 10 as shown in FIG.

FIG. 5 is a flow diagram illustrating one embodiment of a method for a personalized information search server 14 to retrieve information requested from a customer terminal 10.

1, 2 and 5, the customized information search server 14 receives customer information such as a customer ID and a password from the customer terminal 10 (step 100). The personalized information search server 14 confirms that it is a registered customer and receives a search word from the customer terminal 10 (step 105). At this time, the customized information search server 14 may receive the attribute information of the search word from the customer terminal 10 together.

The customized information search server 14 checks whether the information ontology for the search word received from the customer terminal 10 is built in the ontology DB 40 (step 110). Here, the information ontology for the search word is location information in which a document including the search word is registered in the information providing website as described above. In operation 110, if the information ontology is built in the information ontology DB 40, the real-time information retrieval driver 34 of the customized information retrieval server 14 may refer to the information ontology DB 40 to display a document including the corresponding search word. You can search the document by accessing the existing location directly. In addition, the meta search driver 32 of the personalized information search server 14 searches for a document including a search word from the documents registered in the information providing sites before the search point (step 125), and retrieves the searched document from the search information DB. In operation 14b, the controller 14 stores the stored information at 14b.

On the other hand, if the information ontology has not been established in step 110, by accessing the URLs of the information providing websites in real time to search the entire document registered in the information providing websites (step 115), and analyze the location structure of the retrieved documents. The information ontology DB 40 is updated (step 120), and the documents retrieved in step 115 are stored in the search information DB 14b (step 130).

After operation 130, the document stored in the search information DB is analyzed to remove garbage documents to extract useful documents, weight the extracted useful documents (step 135), and the weighted useful documents through a communication network. Provided in real time to the customer terminal (step 140).

FIG. 6 is a flowchart illustrating a weighting step (step 135) in FIG.

Referring to FIGS. 2 and 6, the document editing unit 92 removes stopwords, such as surveys, from a document stored in the search information DB 14b, and uses obsolete garbage by using the attribute information with reference to the knowledge base DB 60. The information is removed to extract only documents useful to the customer (step 135a). After step 135a, the document editing unit 92 re-edits the document by classifying the extracted document by attribute (alias, region, keyword, hint ...) using the attribute information stored in the knowledge base DB 60. (Step 135b), the position of the extracted document is extracted (step 135c).

After step 135c, the weight assigning unit 94 weights each document by using the result and position information analyzed according to the attribute information in the document editing unit 92 (step 135d). As described above, the location of the extracted document is information reflecting its importance, and how many words representing the attribute information of the search word and the search word stored in the knowledge base DB are included in the content of the extracted document and are related to each other. The weight is determined based on whether or not it appears. In addition, a weight may be given to each attribute information. That is, even if the word corresponding to the attribute information appears in the extracted document the same number of times, the weight of the extracted document may vary according to the weight of the attribute information. In addition, when words corresponding to high-weighted attribute information are grouped together in the extracted document, the weight is also increased. In this way, the extracted document is analyzed according to the location and attribute information of the document, and the priority of the extracted document is determined.

After step 135d, the priority of the document is determined according to the assigned weight and stored in the document DB 96 (step 135e).

On the other hand, the above description has been given of a method of searching for information in real time when a search is requested from the customer terminal 10 and providing the searched result to the customer terminal 10. However, as described above, the personalized information search server 14 may be linked to homepages of specific persons, institutions, companies, and schools. In this case, referring to FIGS. 1 and 2, the personalized information search server 14 constructs an information ontology for the search word by using the linked person, institution, company, and school name as the search word. Then, the document is searched periodically by accessing the information ontology directly from the information providing websites to the location where the information about the search word is registered, and storing the searched document in the search information DB 14b.

Then, by using the attribute information stored in the knowledge base DB (60) to remove garbage documents that were incorrectly retrieved from the document stored in the search information DB (14b) to extract only useful documents, using the location information and attribute information of the extracted document By weighting each document, the priority of the documents is determined according to the weights, and stored in the document DB 96. As such, while storing the weighted documents in the document DB 96, when the customer accesses the corresponding homepage and requests related information, the personalized information search server 14 receives the weighted information stored in the document DB 96. The document is extracted and provided to the customer terminal 10.

On the other hand, the present invention can also be embodied as computer readable codes on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like, and may also be implemented in the form of a carrier wave (for example, transmission over the Internet). Include. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

As described above, the customized information retrieval system and method thereof according to the present invention can pre-construct the web page structure of websites providing information as an information ontology to speed up the real-time search, and to search the search terms that the customer wants to search. Specifically, since the attribute information that can be limited is built in advance as a knowledge base DB, garbage information retrieval can be minimized. In addition, weights are assigned using the attribute information and the location information of the retrieved document, and the customer can obtain information suitable for his / her needs by selecting a document with high weight.

Claims (10)

In the knowledge-based customized information retrieval system for providing a search result to the customer terminal in response to a search request from a customer terminal connected through a communication network, Receiving a search request including a search word and attribute information defining the search word from the customer terminal, and accessing a plurality of websites that provide information when the search request is received from the customer terminal or periodically including the search word; An information collection unit for searching documents and constructing a web page structure of websites as an information ontology; A search information database in which documents searched by the information collecting unit are stored; And Based on the attribute information, useful documents are extracted from the searched documents, and priority of the extracted documents is determined based on differentiated weights among attribute words constituting the attribute information set through a knowledge-based method. Customized information system for sorting the extracted documents according to the priority and providing them to the customer terminal or providing the extracted documents to the customer terminal sorted according to the determined priority according to the information providing request received from the customer terminal. Knowledge-based customized information retrieval system, including; studying. delete The method of claim 1, wherein the information collecting unit A URL list storage unit for storing URL information of each of the information providing web sites; Information ontology database; Search terms and attribute information requested to be searched from the customer terminal are received through a communication network or periodically, a document including a search word is extracted using a search system operated by each of the information providing web sites, and the information ontology data is also extracted. A search driver that directly accesses a location of a document for the search word by referring to a base, extracts a real-time document, and stores the extracted document in the search information database; And A knowledge-based customized information retrieval system, comprising: an information ontology constructing unit for analyzing a web page structure that is location information of a searched document stored in the search information database and storing the analyzed result as an information ontology in the information ontology database . According to claim 1, Custom information providing unit A knowledge base database in which search terms frequently appearing in an information providing site and related words are stored as attribute information; And And a knowledge base weighting unit for assigning a weight to each document stored in the search information database using a search term, an association between the attribute information extracted using a knowledge base method and a document, and an importance of each attribute information. Knowledge-based customized information retrieval system. The method of claim 4, wherein the weighting unit A document editing unit which removes a stop word such as a search from a searched document stored in the search information database and removes a garbage document using search terms and attribute information; A weighting unit which removes garbage documents by the document editing unit, and then weights each document using search terms, attribute information, document position information, importance of attribute information, and the degree of relevance between attributes; And And a document database storing a document weighted by the weighting unit. In the knowledge-based customized information service method for providing a search result to the customer terminal in response to a search request from a customer terminal connected through a communication network, (a) receiving a search request including a search word and attribute information defining the search word from the customer terminal; (b) determining whether an information ontology for the search word is established; (c) If the information ontology is established, the information ontology is directly accessed from the information providing websites to the location where the information corresponding to the search word is registered, and the search is performed in real time. Storing; (d) extracting a document to be provided to the customer terminal by removing garbage documents from the documents stored in the search information database corresponding to the search word based on the importance of the attribute information set through the knowledge base method, and extracting the knowledge Determining the priority of the extracted document based on the weights assigned between the attribute words constituting the attribute information through the based method and providing the extracted document to the customer terminal; (e) accessing a URL of each of the information providing websites if the information ontology is not established, searching for the information providing website in real time based on the search word, and storing the information providing website in the search information database; And and (f) analyzing the information structure of the document obtained by searching the information providing website in real time and constructing an information ontology for the search word. delete The method of claim 6, wherein step (d) (d1) removing the stopwords such as the search and removing the garbage information using the attribute information related to the search word; (d2) extracting location information of the retrieved document; (d3) generating a weight of each searched document using search words, attribute information, and location information; And and (d4) determining the priority of the document according to the weight generated in the step (d3). delete delete