KR100312430B1 - System for learning information of goods in internet shopping malls and method using the same - Google Patents

Abstract

The present invention relates to an Internet shopping mall product information learning system and a method thereof, and a computer-readable recording medium recording a program for realizing the same. The query template generating means receives a URL of a new shopping mall from the outside, and generates a query template. Generating a first step; A second step of classifying the search result form of the new shopping mall by using a test query in response to the search result classification means receiving the query template from the query template generating means; A third step of the learning means extracting the position and pattern of the product information according to the search result classification of the search result classification means; And a fourth step of storing the query template generated by the query template automatic generation means, the location of the product information extracted from the learning means, and the pattern of the product information in the knowledge base means.

Description

Internet shopping mall product information learning system and its method {SYSTEM FOR LEARNING INFORMATION OF GOODS IN INTERNET SHOPPING MALLS AND METHOD USING THE SAME}

The present invention relates to an internet shopping mall product information learning method and a computer-readable recording medium recording a program for realizing the same.

In recent years, the Internet has become commonplace, and the number of users who shop using the Internet shopping mall has surged. However, since each shopping mall has a unique search process and a method of outputting a search result, a necessity of a comparative search method for a corresponding product has emerged in an internet shopping mall for delivering information of comparison shopping to a user.

Accordingly, in the related art, a manager who manages a shopping mall has a scheme of directly configuring and storing a rule for extracting elements (product name, manufacturer, product price, etc.) necessary for integration in each shopping mall. The method of collecting and processing the product information in each shopping mall and providing the result to the user based on the manually written rules was used.

However, in the conventional Internet shopping mall, the product information learning method requires a lot of time and manpower to collect and process the product information required by the user, and each time a new shopping mall is created, a new information extraction rule for the shopping mall must be created. There is.

In addition, the shopping mall integrated software called Shopbot is conventionally used, but since the shopbot sets a bias that places a restriction on the system usage environment (a separator must exist for each product information unit to distinguish product information). However, there is a problem that learning about product information of a shopping mall without a delimiter by information unit cannot be performed.

Therefore, the present invention devised to solve the above problems, by automatically generating the necessary product information extraction rules for the comparison shopping for the sale products in the Internet shopping mall, the information required in each shopping mall without the administrator's intervention An object of the present invention is to provide a method for learning product information that can be extracted from an internet shopping mall and a computer-readable recording medium that records a program for realizing the same.

Figure 1 is an embodiment block diagram showing a schematic configuration of an Internet shopping mall product information learning system to which the present invention is applied.

Figure 2 is an embodiment of the processing flow diagram of the Internet shopping mall product information learning method according to the present invention.

FIG. 3 is a detailed flowchart of one embodiment of the query template generation subroutine of FIG. 2. FIG.

Figure 4 is a detailed flowchart of one embodiment of the search result classification subroutine of Figure 2;

FIG. 5 is a detailed flowchart of an embodiment of a location extraction subroutine of the product information of FIG. 2; FIG.

FIG. 6 is a detailed flowchart of an embodiment of a pattern extraction subroutine of product information in FIG. 2; FIG.

* Explanation of symbols for the main parts of the drawings

10: query template generator 12: search result classifier

14: table type learner 16: knowledge base

18: list type learner

Internet shopping mall product information learning system to which the present invention is applied to achieve the above object, Query template generation means for generating a query template by receiving a newly generated Uniform Resource Locator (URL) of the shopping mall; Search result classification means for receiving a query template from the query template generating means and receiving a test query word to classify a search result form of a new shopping mall; Learning means for extracting product location and pattern information from a search result page of a shopping mall classified by the search result classification means; And a knowledge base means for storing the query template generated by the query template generating means and the position information and the pattern information of the product extracted by the learning means.

In accordance with another aspect of the present invention, there is provided a method for learning product information on an internet shopping mall, the query template generating means receiving a URL of a new shopping mall from the outside to generate a query template; A second step of classifying the search result form of the new shopping mall by using a test query in response to the search result classification means receiving the query template from the query template generating means; A third step of the learning means extracting the position and pattern of the product information according to the search result classification of the search result classification means; And a fourth step of storing the query template generated by the query template automatic generation means, the location of the product information extracted from the learning means, and the pattern of the product information in the knowledge base means.

In addition, the present invention, the Internet shopping mall product information learning system having a microprocessor, the query template generating means for receiving a URL of the new shopping mall from the outside, the first function to generate a query template; A second function of the search result classification means receiving the query template from the query template generating means to classify the search result form of the new shopping mall by using a test query; A third function of the learning means extracting the position and pattern of the product information according to the search result classification of the search result classification means; And a program for storing a query template generated by the query template automatic generation means, a fourth function of storing the location of the product information extracted from the learning means and the pattern of the product information in the knowledge base means. Provide a record carrier.

The above objects, features and advantages will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Since the Internet shopping mall has its own unique search processing and output method of search results, it can have a fully structured search result form and a semi-structured search result form. In this case, the fully structured search result refers to a case in which there is a label indicating what information is included in each column, such as a table in a database, and the semi-structured search result refers to a form in which a pattern is regularly repeated.

The present invention proposes an internet shopping mall product information learning method that can automatically extract information not only from a fully structured search result but also from a semi-structured search result.

1 is a block diagram of an embodiment of a schematic configuration of an Internet shopping mall product information learning system to which the present invention is applied.

As shown in FIG. 1, a query template generator 10 for generating a query template by receiving a Uniform Resource Locator (URL) of a newly created shopping mall, and receiving and testing a query template from the query template generator 10. The search result classifier 12 which classifies the search result form of the new shopping mall by receiving a query word, and when the search result type classified by the search result classifier 12 is a table type, the location information of the product is displayed on the search result page of the shopping mall. A table-type learner 14 for extracting and a list-type learner 16 for extracting pattern information of a product from a search result page of a shopping mall when the search result type classified by the search result classifier 12 is a list type; Knowledge base (Knowledge-base) for storing the location information and pattern information of the query template and the product generated by the query template generator 10 ( 18). Here, the table type refers to a form in which product information is output in the form of a database using a <TABLE> tag of HTML (HyperText Markup Language), and the list type refers to the form of all search results other than the table type.

Next, a method of learning online shopping mall product information to which the present invention is applied will be described with reference to FIGS. 2 to 6.

2 is a flowchart illustrating an embodiment of a method for learning an internet shopping mall to which the present invention is applied. When a new shopping mall is created, a URL (Uniform Resource Locator) of the shopping mall is received (100), and the URL of the received shopping mall is received. When the query template generator 10 is input to the query template generator 10, the query template generator 10 generates a query template 102 and transmits the generated query template to the search result classifier 12.

The search result classifier 12 receiving the query template from the query template generator 10 classifies the search result form of the new shopping mall by using the test query (104).

When the search result classified by the search result classifier 12 is a table type, the table type learner 14 extracts the location of the product information from the shopping mall page (106) and classifies it in the search result classifier 12. If the search result is a list type, the list type learner 16 extracts a pattern of product information (108).

The query template generated by the query template automatic generator 10, the location of the product information extracted by the table type learner 14, and the pattern of the product information extracted by the list type learner 16 may be used in the knowledge base 18. Stored at 110.

Attached FIG. 3 is a detailed flowchart of a process 102 of generating a query template by the query template generator 10 in FIG. 2, and when a user inputs a search term for a desired product in an offline operation, In order to query each shopping mall, it is used to determine the query form of each shopping mall in advance.

First, an HTML (HyperText Markup Language) file of a shopping mall to be learned through a webrobot is collected, and the number i of <FORM> tags used for searching is obtained through HTML parsing (200). It is determined whether the number i of the <FORM> tags is 1 (202).

In response to the determination result 202, when the number i of the <FORM> tags is not 1, the query template generator 10 may input the number j of the <INPUT> tags whose type is TEXT to input a search keyword. In operation 204, it is determined whether the number j of the <INPUT> tags is 1 (206). Here, a general method for passing the search keyword as a parameter to a CGI (Common Gateway Interface) program is to use an <INPUT> tag having a tag type of text (TEXT). This is because the number of <INPUT> tags of TEXT type may exist.

As a result of the determination 202, when the number i of the <FORM> tags is 1, the query template generator 10 selects a <FORM> tag suitable for performing a search among a plurality of <FORM> tags. In general, a number of <FORM> tags may exist on a page of an HTML document, which means that a CGI program of a shopping mall server can be run through several paths on a page. Therefore, <FORM> is selected to actually search.

In the determination result 206, when the number j of the <INPUT> tags is 1, the query template generator 10 generates a query template by combining a method of transmitting a query inside the <FORM> tag to the shopping mall server ( 208).

In response to the determination result 206, when the number j of the <INPUT> tags is not 1, the query template generator 10 selects an <INPUT> tag for inserting a query word into any <INPUT> tag among a plurality of <INPUT> tags. Select 210, and perform step 208.

4 is a detailed flowchart of a process 104 in which the search result classifier 12 receiving the query template from the query template generator 10 in FIG. 2 classifies a search result form of a new shopping mall using a test query. The search result classifier 12 receives the query template from the query template generator 10 and determines whether a <TABLE> tag exists in the HTML file storing the new shopping mall search result (300).

In the determination result 300, if the <TABLE> tag exists, the search result classifier 12 separates a portion starting with <TABLE> and ending with </ TABLE> from an HTML file storing a new shopping mall search result. In operation 302, if the determination result 300 does not include a <TABLE> tag, the search result classifier 12 sets the type of the search result of the current shopping mall to a list type, and the new shopping mall search result. Transfer the stored HTML file to the list type learner (16).

After performing step 302, the search result classifier 12 determines whether the label and ontology of the <TABLE> tag are the same in the HTML file storing the new shopping mall search result, and the product information is repeatedly displayed in the same column. (304) Here, the <TABLE> tag in the HTML file containing the search results of the new shopping mall cannot be confirmed as part of the product information. Therefore, it is necessary to discriminate based on the keywords in the first row of the current table and the contents of the table. do. In general, the first line of the <TABLE> tag of the HTML file that stores the search results of a shopping mall has a label indicating what information the column contains. Ontology can be viewed as a kind of dictionary used in e-commerce. In other words, the words 'commodity price', 'Price', 'sale price' and the like correspond to the ontology when referring to the product price.

If the determination 304, the label of the <TABLE> tag and the ontology is the same, and the product information repeatedly appears in the same column, the search result classifier 12 is a table of the type of the search results of the current shopping mall Type, the HTML file storing the new shopping mall search result is sent to the table type learner 14 (306), and the determination result 304, the label of the <TABLE> tag and the ontology are the same. When the product information does not repeatedly appear in the same column, the search result classifier 12 sets the form of the search result of the current shopping mall to a list type, and displays the HTML file storing the new shopping mall search result in the list type learner 16. To send.

5 is a detailed flowchart of a process 106 of extracting the location of the product information from the shopping mall page by the table type learner 14 in FIG. 2, and the search result of the new shopping mall is searched from the search result classifier 12. Receiving the stored HTML file, the table type learner 14 generates information between the <TABLE> tag and the </ TABLE> tag of the current page into a separate file (400), and to find the location information of the product. Retrieve the first row of the table (402).

The table-type learner 14 determines whether the keywords appearing in the tag <TD> match the ontology for all the <TD> tags of the HTML file in order to determine the location of the information to be extracted from the table of the HTML file. (404).

In the determination result 404, when the keywords appearing in the tag <TD> and ontology match, the table-type learner 14 recognizes the positions of the columns necessary for the integration and makes the positions in the form of a rule to display the product information. Information about the location is registered (406).

Here, when the product search result is a table type, other columns of the same row indicate information about one product. Each column also represents the same semantic information for different products. For example, if one column represents price information, each cell included in the column represents price information of another product.

The first row of the table shows the meaning of the information represented by each column in the form of keywords. If a column contains price information, keywords such as 'price, consumer, sale price, price' will appear.

6 is a detailed flowchart of the process 108 of extracting the pattern of the product information by the list type learner 16 in FIG. 2, and the list type learner 16 receives the new result received from the search result classifier 12. To divide the shopping mall search result HTML file into product information units, divide the line by line by using the <BR> tag and line break tag of HTML, and save the shopping mall page to be analyzed for every line of the HTML file. The line is read (500), the meaning of the information contained in the current line is identified (502), the meaning is recorded (504), and the product information position pattern displayed on the page is extracted (508).

For example, in the case of a book, if the information to be extracted is 'book name, price', the book name may be checked by matching whether the query word used to request a search from a shopping mall server is included. In addition, to determine whether the current line represents a price, it is necessary to check whether numeric information appears after a keyword indicating a price, that is, a price, a selling price, a price, and the like.

That is, the pattern is extracted to accurately extract the product information unit for one product. The repetitive pattern is product information representing one product. When a unit of product information is extracted, only necessary product attributes are extracted from the product information to integrate the information.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit of the present invention for those skilled in the art to which the present invention pertains. It is not limited to the drawing.

The present invention as described above, by automatically generating the information extraction rules of the new shopping mall can reduce the manual labor of the administrator, there is an effect that can easily maintain the comparison shopping system,

In addition, when information cannot be extracted through the rules already created, that is, when the search method of the existing shopping mall and the output method of the search result are different, it is automatically detected and updated without any additional work by the administrator. The effect is to significantly reduce the cost required to maintain the system.

Claims (11)

Query template generating means for receiving a newly created URL (Uniform Resource Locator) of the shopping mall and generating a query template; Search result classification means for receiving a query template from the query template generating means and receiving a test query word to classify a search result form of a new shopping mall; Learning means for extracting product location and pattern information from a search result page of a shopping mall classified by the search result classification means; And Knowledge base means for storing location information and pattern information of the query template generated by the query template generating means and the product extracted by the learning means Internet shopping mall product information learning system comprising a. The method of claim 1, The search result classifying means, the Internet shopping mall product information learning system characterized in that it can automatically classify the search result form of the new shopping mall in the form of a table or list. The method of claim 1, The learning means, Table type learning means for extracting location information of a product from a table type search result page classified by the search result classification means; And Internet shopping mall product information learning system including a list-type learning means for extracting the pattern information of the product from the list type search results page classified by the search result classification means. The method of claim 3, wherein The list-type learning means, the Internet shopping mall product information learning system, characterized in that to divide the new shopping mall search results HTML file received from the search result classification means line by line using the <BR> tag and the line break tag of the HTML. A first step of generating, by the query template generating means, a URL of a new shopping mall from the outside to generate a query template; A second step of classifying the search result form of the new shopping mall by using a test query in response to the search result classification means receiving the query template from the query template generating means; A third step of the learning means extracting the position and pattern of the product information according to the search result classification of the search result classification means; And A fourth step of storing the query template generated by the query template automatic generation means, the location of the product information extracted from the learning means, and the pattern of the product information in the knowledge base means; Internet shopping mall product information learning method comprising a. The method of claim 5, The first step, A fifth step of collecting, by the query template generating means, an HTML file of a shopping mall to be learned, obtaining the number of <FORM> tags used for searching, and determining whether the number of <FORM> tags is 1; A sixth step of, if the number of <FORM> tags is 1, the query template generating means obtains the number of <INPUT> tags and determines whether the number of <INPUT> tags is 1; A seventh step of selecting, by the query template generating means, a <FORM> tag suitable for performing a search among a plurality of <FORM> tags when the number of <FORM> tags is 1 as a result of the determination in the fifth step; An eighth step of generating, by the query template generating means, the query template when the number of <INPUT> tags is 1 as a result of the determination of the sixth step; And If the number of <INPUT> tags is not 1 as the determination result of the sixth step, the query template generating means includes a ninth step of selecting an <INPUT> tag into which the query word is to be inserted and performing the eighth step. Internet shopping mall product information learning method. The method according to claim 5 or 6, The second step, A tenth step of determining, by the search result classification means that receives the query template from the query template generating means, whether a <TABLE> tag exists in an HTML file storing a new shopping mall search result; As a result of the determination in the tenth step, if the <TABLE> tag is present, the search result classification means stores a portion starting with <TABLE> and ending with </ TABLE> in an HTML file in which the new shopping mall search result is stored as a separate file. An eleventh step of generating; As a result of the determination in the tenth step, if the <TABLE> tag does not exist, the search result classification means sets the form of the search result of the current shopping mall to a list type, and lists the HTML file storing the new shopping mall search result. A twelfth step of transmitting to a learning means; After performing the eleventh step, the search result classification means determines whether the label and the ontology of the <TABLE> tag are the same in the HTML file storing the new shopping mall search result, and the product information repeatedly appears in the same column. The thirteenth step; If the label and the ontology of the <TABLE> tag are the same and product information repeatedly appears in the same column as the result of the determination of the thirteenth step, the search result classification means sets the form of the search result of the current shopping mall to a table type, A fourteenth step of transmitting the HTML file storing the new shopping mall search results to the table type learning means; If the label and ontology of the <TABLE> tag are the same and product information does not repeatedly appear in the same column as a result of the determination of the thirteenth step, the Internet shopping mall product information learning method comprising the fifteenth step of performing the twelfth step . The method of claim 7, wherein The third step, A sixteenth step in which the table-type learning means extracts the location information of the product from the search result page of the shopping mall when the search result type classified by the search result classification means is a table type; And And when the search result type classified by the search result classification means is a list type, the list type learning means comprising a seventeenth step of extracting pattern information of a product from a search result page of the shopping mall. The method of claim 8, The sixteenth step, An eighteenth step in which the table-type learning means receiving the HTML file storing the search results of the new shopping mall from the search result classification means generates information between the <TABLE> tag and the </ TABLE> tag of the current page in a separate file; step; A nineteenth step, wherein the table-type learning means searches a first row of a table to find location information of a product; A twenty step of determining, by the table-type learning means, for all <TD> tags in the HTML file, whether keywords and ontologies that appear in a tag <TD> match; And As a result of the determination in step 20, when the keywords appearing in the tag <TD> and ontology coincide, the table-type learning means recognizes the positions of the columns necessary for the integration and makes the positions in the form of a rule to position the product information. Internet shopping mall product information learning method comprising the step 21 of registering information about. The method according to claim 8 or 9, The seventeenth step, A twenty-second step of receiving, by the list type learning means, a new shopping mall search result HTML file from the search result classification means, dividing the HTML file line by line using a tag and a line break tag; A twenty-third step in which the list-type learning means reads one line from a file storing a shopping mall page to be analyzed for every line of the HTML file, and grasps the meaning of the information contained in the current line; And And a twenty-fourth step, wherein the list-type learning means extracts the product information position pattern shown in the HTML file. In an online shopping mall product information learning system having a microprocessor, A first function of the query template generating means receiving a URL of a new shopping mall from the outside to generate a query template; A second function of the search result classification means receiving the query template from the query template generating means to classify the search result form of the new shopping mall by using a test query; A third function of the learning means extracting the position and pattern of the product information according to the search result classification of the search result classification means; And A fourth function for storing the query template generated by the query template automatic generation means, the location of the product information extracted from the learning means, and the pattern of the product information in the knowledge base means A computer-readable recording medium having recorded thereon a program for realizing this.