KR100921894B1 - Method and system for classifying documents using classification information of query - Google Patents

Abstract

PURPOSE: A method and system for classifying documents using classification information of query are provided to determine the intention of the user reflected to a query exactly and easily. CONSTITUTION: A method for classifying documents using classification information of query includes the step of maintaining the query database(310) which is stored by matching the individual category information of a query with the corresponding query; the step of determining the individual category information of the corresponding query as the individual category information of the word in case the query corresponding to the word included in the object of classification document is included in the query database; the step of producing the category information of the object of classification document.

Description

Method and System for Classifying Documents Using Classification Information of Query}

The present invention relates to document classification, and more particularly, to a method of classifying a document using classification information of a query input by a user.

Due to the development and spread of the Internet, various services using the Internet are provided, and a representative example thereof is a search service. When a search service inputs a word or a combination of words to be searched by a user, the search engine searches a search result document corresponding to the search engine (eg, a web site, an article, or the search query input from the user). A service that provides a user with an image having a file name including the corresponding search query).

As can be seen above, the query input by the user to use the search service includes the search intention of the user. Therefore, if the user can accurately analyze the intention that the user entered the query, not only can efficiently provide various contents in relation to a search service, but can also extend the scope of providing contents using the Internet.

One of the typical methods of analyzing the intentions of users entering a query is to use query classification. However, as described above, since the query input by the user is a single word or a combination of only a few words, it is not easy to correctly classify the query.

Meanwhile, in order to provide search result documents classified by category through the above-described search service or to provide documents classified on the Internet by category, search result documents or documents on the Internet (hereinafter referred to as 'web documents') are classified into categories. Work is required. However, in the related art, since the classification of the web document is performed by manual, there is a problem that the classification takes a lot of cost and time.

The present invention is to solve the above-described problem, query classification that can generate the classification information for each category of the query by using the classification information of the search result document for the query or the category information included in the search result document for the query It is a technical problem to provide a method and system for generating information.

Another object of the present invention is to provide a document classification method and a system capable of classifying documents using document classification information generated by using category information for each category of a query.

According to an aspect of the present invention, there is provided a document classification method using query classification information. Determining classification information for each category of the query as classification information for each category of the word when a query corresponding to a word included in a document to be classified is included in the query database; And generating classification information of the classification target document by using classification information for each category of words included in the classification target document.

In one embodiment, in the document classification information generation step, the document classification information is generated by summing classification information for each category of words included in the classification target document for each category, and generates the generated document classification information in percentage units. The document classification information is generated by the conversion.

Meanwhile, the document classification method using the above-described query classification information may further include generating classification information for each category of the query. In one embodiment, the classification information for each category of the query is generated by summing the category information for each category of search result documents for the query for each category, wherein the classification information for each category of the search result documents is a word construction method or Characterized by the document learning method. In another embodiment, classification information for each category of the query is generated by summing category information included in a search result document for the query for each category, wherein the search result document is one of the search result documents for the query. And a search result document selected by the users.

On the other hand, the document classification method using the above-described query classification information further comprises the step of providing the search result documents corresponding to the search request classified by category using the document classification information when a search request is received from a user It features.

According to another aspect of the present invention, there is provided a document classification system using query classification information. Word classification information determination unit which extracts words included in a document to be classified, and determines classification information for each category of the query as classification information for each category of the word when a query corresponding to the extracted word is included in the query database. ; And a document classification information generation unit configured to generate classification information of the document to be classified, using classification information for each category of each word determined by the word classification information determination unit.

According to an aspect of the present invention, there is provided a method of generating query classification information, the method comprising: obtaining classification information for each category of search result documents for a query; And generating category information of each category of the query by adding category information of each category of the search result document by category. In this case, the classification information for each category of the search result documents may be generated using word construction or document learning.

In accordance with another aspect of the present invention, there is provided a method of generating query classification information, comprising: extracting category information included in each search result document from a search result document for a query; And generating classification information for each category of the query by adding the extracted category information for each category. At this time, the search result document for the query is characterized in that the search result document selected by the user.

In accordance with still another aspect of the present invention, there is provided a system for generating query classification information, comprising: a category information extracting unit for extracting category information from each search result document for a query; And an operation unit configured to generate classification information for each category of the query by adding the extracted category information for each category.

As described above, according to the present invention, since the classification information for each category of the query is generated using the classification information of the search result document for the query or the category information included in the search result document for the query, the intention of the user reflected in the query is reflected. Can be more accurately and easily determined, and the scope of content provision through the Internet can be extended by using such query classification information.

In addition, the present invention can generate the document classification information by using the classification information for each category of the query containing the user's input intent, and can classify the document by using the document classification information, as well as maintaining the latest document classification, The automation of document classification can be secured, and the automation of such document classification can reduce the time and effort required for document classification.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a schematic block diagram of a system for generating query classification information according to an embodiment of the present invention. As shown, the query classification information generation system 100 includes a document classification information DB 110, a document classification information extracting unit 112, and a first calculating unit 114.

The document classification information DB 110 stores classification information for each category of documents. Here, the document classification information means scoring the degree to which the corresponding document belongs to a specific category. For example, if the degree to which a particular document belongs to the category of drama is 0.57 and the degree to belong to the category of history is 0.33, the document classification information of the document is defined as "Drama: 0.57", "History: 0.33". The document classification information DB 112 stores the document classification information in correspondence with each document.

In one embodiment, document classification information of each document may be calculated using a document learning method or a word construction method. Hereinafter, a method of generating document classification information using a document learning method and a word construction method will be described in detail.

In the document learning method, first, a predetermined number of documents selected by users are classified into categories, and words included in each document are extracted from each document, and the category information to which the document belongs is designated in the extracted word. . This process is repeated for the documents included in all the categories, and then the category classification of each word is determined by summing the number for each category specified in each word, and the word database is constructed using this information.

Subsequently, when a document classification is requested for a specific document, a word included in the document is extracted, and if the extracted word is stored in the word database, the category information matching the word is determined as the category information of the extracted word. In addition, by summing the category information of all the words included in the specific document for each category, document classification information of the specific document is generated.

Meanwhile, in the word construction method, first, representative words belonging to each category are selected, and the same category information as the representative words for other words that appear together with the representative words in the document for a predetermined number or more of documents. Specifies. This process is repeated for all documents, and then the category classification of each word is determined by summing the number for each category assigned to each word, and a word database is constructed using this information.

Subsequently, when a document classification is requested for a specific document, a word included in the document is extracted, and if the extracted word is stored in the word database, the category information matching the word is determined as the category information of the extracted word. In addition, by summing the category information of all the words included in the specific document for each category, document classification information of the specific document is generated.

The document classification information extracting unit 112 extracts document classification information of each search result document for the query from the document classification information DB 110. Specifically, document classification information of search result documents generated by performing a search for a predetermined query is extracted from the document classification information DB 110. For example, there are two search result documents for the query " King and I ", and the document classification information of the first search result document is "Drama: 0.57", "History: 0.33", and the document of the second search result document. When the classification information is "movie: 0.21", drama: 0.55 ", the document classification information extracting unit 112 extracts document classification information of each of these search result documents from the document classification information DB 110.

The first calculation unit 114 generates the classification information for each category of the query by summing the document classification information of the respective search result documents extracted by the document classification information extraction unit 112 for each category. For example, as in the above example, in the case of the query "Wang and Me", the document classification information of the first search result document is "Drama: 0.57", "History: 0.33", and the document classification information of the second search result document is Since "movie: 0.21" and "drama: 0.55", the document classification information is added to each category to "drama: 1.12", "history: 0.33", and "movie: 0.21". Therefore, the categorization information for each category of the query "Wang and Me" is "Drama: 1.12", "History: 0.33", and "Movie: 0.21".

The query classification information generation system 100 may manage and classify the classification information for each query generated through this process.

2 is a schematic block diagram of a system for generating query classification information according to another embodiment of the present invention. As shown, the query classification information generation system 200 includes a category information extraction unit 210, a query-specific selection document DB 212, and a second operation unit 214.

The category information extracting unit 210 extracts category information included in each search result document for a predetermined query from each search result document. Specifically, after providing a search service, a search result document provided through a search collection such as an intellectual person includes category information including an ID or a category name of a category to which the search result document belongs in the search result document. The extraction unit 210 extracts such category information from each search result document.

In one embodiment, the category information extractor 210 may extract category information from all search result documents for a given query, but in a modified embodiment, the category information extractor 210 may only category the search result documents selected by users for the query. Information can be extracted. To this end, the query classification information generation system 200 may match the search result documents selected by the users among the search result documents for the query with the corresponding query and store the selected query DBs 212 for each query.

The second calculating unit 214 generates category information for each category of the corresponding query by summing the category information extracted by the category information extracting unit 210 for each category. In this case, the category information may be extracted from the search result document selected by the users of the search result document for the query as described above, the second operation unit 214 is the result value obtained by summing the extracted category information for each category By converting in percentage units, you can generate classification information for each category of the query.

For example, if there are 10 documents selected by the users for the query "King and Me", the category information extracted from the first search result document and the fifth search result document is drama, and the second to fourth search result documents and the first search result document are selected. If the category information extracted from the sixth to eighth search result documents is history, and the category information extracted from the ninth and tenth search result documents is a movie, when each category information is added by category, there are two dramas, six histories, and a movie. Are two, so if you convert these values in percentage, you can see that the categorization information for the query "Wang and Me" is defined as "History: 0.6", "Drama: 0.2", and "Movie: 0.2". .

The query classification information generation system 200 may manage and classify the classification information for each query generated through this process.

As described above, the system for generating query classification information according to an embodiment of the present invention generates query classification information by using document classification information of a search result document for a corresponding query, and uses the query classification information to classify each query by category. Because it can be classified, the user's intention reflected in the query can be judged more accurately and easily.

3 is a schematic block diagram of a document classification system capable of classifying documents using query classification information. As shown, the document classification system 300 includes a query DB 310, a word classification information determiner 320, and a document classification information generator 330.

The query DB 310 stores the classification information of each query by matching it with the corresponding query. In one embodiment, the classification information for each category of the query may be generated by the above-described query classification information generation system (100, 200).

In the present embodiment, the query DB 310 is described as being included in the document classification system 300, but the query DB 310 is included in the above-described query classification information generation system 100 and 200, and the document classification system ( The document classification system 300 does not directly include the query DB 310 in the modified embodiment, since 300 may only receive classification information for each query category from the query classification information generation systems 100 and 200. You may not.

In addition, in the above-described embodiment, the document classification system 300 uses classification information for each category of each query provided from the query classification information generation systems 100 and 200 which are separate components from the document classification system 300. Although described, in the modified embodiment, the document classification system 300 may directly include the query information generation system in one module form.

The word classification information determiner 320 receives a classification target document, extracts words included in the classification target document from the classification target document, and determines classification information for each category of the corresponding words.

In one embodiment, the word classification information determiner 320 determines whether the same query as the word extracted in the query DB 310 exists, and if present, the classification information for each category of the corresponding query is extracted. Can be determined by the category information by category.

The document classification information generation unit 330 generates document classification information for each category of the document to be classified using the classification information for each category of each word determined by the word classification information determination unit 320. In one embodiment, the document classification information generation unit 330 may generate document classification information of the corresponding document by summing category information for each category of words included in the document to be classified for each category.

For example, the words "wang and me" and the words "discrete" were extracted from the document to be classified, and the category information for each category of the query "king and me" stored in the query DB 310 is "drama: 1.12", If the "History: 0.33", "Movie: 0.21", and the classification information by category for the query "Discrete" is "Drama: 1.77", "History: 0.53", the classification information of each word category is summed by category Then, the classification information for the category called drama becomes "2.19", the classification information for the category called history becomes "0.86", and the classification information for the category called movie becomes "0.21". Accordingly, the document classification information generation unit 330 generates document classification information "Drama: 2.19", "History: 0.86", and "Movie: 0.21" for the document to be classified.

In one embodiment, the document classification information generation unit 330 may convert the above-described document classification information in percentage units to more accurately analyze which category the document to be classified belongs to.

The document classification information generation unit 330 matches the document classification information of each document generated according to the above-described method with each document and stores the document classification information DB 340.

As described above, the present invention can classify each document using the classification information for each category of the query including the user's input intention, thereby maintaining the latest document classification and automatically classifying the document. This will reduce the time and effort required.

On the other hand, the document classification system 300 may further include a document classification unit 350 for classifying each document by category using the document classification information generated by the document classification information generation unit 300 described above. For example, when the document classification unit 350 receives a search request from the user, the document classification unit 350 uses the document classification information generated by the document classification information generation unit 300 to provide a classification result for each category. The search result documents may be classified, or the web documents included in the blog may be classified and provided to the user by using the document classification information generated by the document classification information generation unit 300.

Hereinafter, a method of generating query classification information according to the present invention will be described with reference to FIGS. 4 to 5.

4 is a flowchart illustrating a method of generating query classification information according to an embodiment of the present invention. As shown, first, classification information of each category of search result documents for each query is extracted from the database (operation 400). Here, the classification information for each category of the search result documents means the degree to which the corresponding document belongs to a specific category. The document classification information is matched with the corresponding document and stored in the database.

In one embodiment, document classification information of each document may be calculated using a document learning method or a word construction method. Detailed descriptions of the document learning method and the word construction method are described in detail above, and thus detailed descriptions thereof will be omitted.

Thereafter, the classification information for each category of the query is generated by summing the classification information for each category of the search result documents extracted from the database for each category (operation 410). For example, if there are two search result documents for the query "Wang and I", the document classification information of the first search result document is "Drama: 0.57", "History: 0.33", and the second search result document When the document classification information is " movie: 0.21 " and " drama: 0.55 ", when the respective document classification information is summed by category, it becomes "drama: 1.12", "history: 0.33", and "movie: 0.21". Therefore, the categorization information for each category of the query "Wang and Me" is "Drama: 1.12", "History: 0.33", and "Movie: 0.21".

Thereafter, a query DB is constructed using classification information for each category of the query (step 420). At this time, the query DB is matched and stored in each category and the classification information of the corresponding query, the constructed query DB can be used to generate document classification information.

5 is a flowchart illustrating a method of generating query classification information according to another embodiment of the present invention.

As shown, first, category information included in each search result document is extracted from the search result document for each query (operation 500). In detail, in providing a search service, a search result document provided through a search collection such as an intellectual person includes category information including an ID or a category name of a category to which the search result document belongs in the search result document. This category information is extracted from the document.

In the above-described embodiment, it is described that category information is extracted from all search result documents for each query. However, in the modified embodiment, category information may be extracted only for the search result document selected by the users for the query. To this end, the present invention can match and store the search result documents selected by the users among the search result documents for the query with the corresponding query.

Thereafter, the extracted category information is added to each category to generate classification information for each category of the corresponding query (step 510). In this case, the category information may be extracted from the search result document selected by the users of the search result document for the query as described above, in generating the category information for each category of the query by using the extracted category information You can convert the result of the sum to a percentage unit.

In detail, category information of drama is extracted from the first search result document and the fifth search result document among the ten search result documents in the document selected by the user for the query "King and Me". 4 Assuming that category information called history is extracted from the search result documents and the sixth through eighth search result documents, and that category information called movie is extracted from the ninth and tenth search result documents, each category information is classified by category. In total, we can see that there are two dramas, six histories, and two movies. In this case, when these values are converted in percentage units, the classification information for each category for the query "Wang and Me" is generated as "History: 0.6", "Drama: 0.2", and "Movie: 0.2".

Finally, a query DB is constructed using the classification information for each category of the query (step 520). As described above, in the query DB, classification information for each query and category of the corresponding query are matched and stored, and the constructed query DB may be used to generate document classification information.

Hereinafter, a method of classifying documents using classification information for each category of a query will be described with reference to FIG. 6.

6 is a flowchart showing a document classification method according to an embodiment of the present invention. First, the query DB in which the classification information for each category of each query is mapped and stored with the corresponding query is maintained (step 600). Here, the classification information for each category of the query may be generated using the method illustrated in FIG. 4 or 5.

Thereafter, when the document to be classified is received (operation 610), words included in the document are extracted from the received classification object document (operation 620). Next, it is determined whether the query corresponding to each extracted word is included in the query DB (step 630). If the query corresponding to the extracted word is included in the query DB, the query is classified by category. The information is determined as classification information for each category of the extracted word (step 640).

Then, it is determined whether there is a word to determine the classification information for each category among the extracted words (step 650), and if not, classify the document of the document to be classified using the classification information for each category of the extracted words In operation 660, the controller 110 repeats steps 630 and 640 described above.

In one embodiment, the document classification information may be generated by summing classification information of each category of extracted words for each category. For example, the words "wang and me" and the words "discrete" have been extracted from the document to be classified, and the classification information for each category of the query "king and me" is "Drama: 1.12", " When history: 0.33 "," movie: 0.21 ", and the categorization information for the query" discrete "is" drama: 1.77 "and" history: 0.53 ", the categorization information for the two words is categorized by category. When added together, the drama becomes "2.19", the history becomes "0.86", and the movie becomes "0.21", so the document titled "Drama: 2.19", "History: 0.86", "Movie: 0.21" The classification information is generated.

In addition, in generating the document classification information, the above-described document classification information may be converted in percentage units to more accurately analyze which category the document to be classified belongs to.

On the other hand, although not shown in Figure 6, when a search request is received from the user to use the document classification information described in the search result documents included in the search results in order to provide the search results corresponding to the search request for each category By classifying or classifying web documents included in blogs into respective categories using the above-described document classification information.

The above-mentioned query classification information generation method and document classification method may be implemented in a program form that can be executed using various computer means. In this case, a program for performing the query classification information generation method and document classification method may include a hard disk, It is stored in a computer-readable recording medium such as a CD-ROM, a DVD, a ROM, a RAM, or a flash memory.

Those skilled in the art to which the present invention pertains will understand that the present invention can be implemented in other specific forms without changing the technical spirit or essential features.

Therefore, it is to be understood that the embodiments described above are exemplary in all respects and not restrictive. The scope of the present invention is shown by the following claims rather than the detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention. do.

1 is a schematic block diagram of a system for generating query classification information according to an embodiment of the present invention.

2 is a schematic block diagram of a system for generating query classification information according to another embodiment of the present invention.

3 is a schematic block diagram of a document classification system according to an embodiment of the present invention.

4 is a flowchart showing a method of generating query classification information according to an embodiment of the present invention.

5 is a flowchart showing a method of generating query classification information according to another exemplary embodiment of the present invention.

6 is a flowchart showing a document classification method according to an embodiment of the present invention.

Claims (15)

Maintaining a stored query database in which classification information for each category of a query is matched with a corresponding query; Determining classification information for each category of the query as classification information for each category of the word when a query corresponding to a word included in a document to be classified is included in the query database; And And generating classification information of the classification target document by using classification information for each category of words included in the classification target document. The method of claim 1, wherein in the document classification information generation step, And the document classification information is generated by summing classification information for each category of words included in the classification target document for each category. The method of claim 1, The method of claim 1, further comprising generating classification information for each category of the query. The method of claim 3, The classification information for each category of the query is generated by summing the classification information for each category of the search result documents for the query for each category. The method of claim 4, wherein The classification information for each category of the search result documents is generated by a word construction method or a document learning method. The method of claim 3, The classification information for each category of the query is generated by summing category information included in a search result document for the query for each category. The method of claim 6, And the search result document is a search result document selected by users among search result documents for the query. The method of claim 1, And classifying the search result documents corresponding to the search request into categories by using the document classification information when receiving a search request from a user. A recording medium having recorded thereon a program for performing the method according to any one of claims 1 to 8. A query database in which classification information for each category of a query is matched with the corresponding query and stored; Word classification information determination unit which extracts words included in a document to be classified, and determines classification information for each category of the query as classification information for each category of the word when a query corresponding to the extracted word is included in the query database. ; And And a document classification information generation unit for generating classification information of the document to be classified by using classification information for each category of each word determined by the word classification information determination unit. The method of claim 10, And the document classification information generation unit generates the document classification information by summing the category information for each category of words determined by the word classification information determination unit for each category. The method of claim 10, Document classification system using the query classification information, characterized in that it further comprises a query classification information generation unit for generating the classification information for each category of the query. The method of claim 12, The query classification information generation unit generates the classification information for each category of the query by summing the category information for each category of the search result documents for the query for each category. The method of claim 13, The query classification information generation unit generates a classification information for each category of the search result documents using a word construction method or a document learning method. The method of claim 12, The query classification information generation unit generates the classification information for each category of the query by using the category information included in the search result document selected by the user among the search result documents for the query. One document classification system.

Images