KR20100125682A - Semantic search method and system integrating multiple classification systems - Google Patents

Abstract

The present invention integrates databased resources into different classification systems in various fields, and uses an ontology classification schema to map the database according to the degree of association with other classification concepts in other fields. The present invention relates to a semantic search method and system capable of realizing a recommendation and semantic resource search service of a classification category with improved user accessibility. In the semantic search method according to the present invention, a database corresponding to each classification category is divided into two or more classification systems having a plurality of classification categories, and the resources corresponding to each classification category are databased. An association relationship between the two classification categories is set based on the similarity values of the first classification category of the classification scheme and the second classification category of the second classification scheme. On the basis of this, the search results may be provided by dividing resources belonging to a plurality of classification systems by each classification system or by each classification category.

Description

{Semantic Search Method and System for Associating with Plurality of Classifications}

The present invention relates to a semantic search method and system, and in particular, by using the ontology classification scheme by integrating the database database in different classification schemes in various fields, different classification concepts in different fields. The present invention relates to a semantic search method and system that can implement a semantic resource retrieval service and recommendation of a classification category with improved user accessibility by mapping and mapping the database according to the degree of association.

In today's flood of information, a system that can quickly and accurately search for information that a user wants to find is required in many fields. For example, a lot of information managed by a system that is useful for searching, such as a book search system in a library, an information search system on a web site, a search system for resources managed on a small network connected to a LAN or a storage device of a personal PC. For example, many methods for retrieving various types of information such as document files, image files, video files, database information, and text data in a faster and more intelligent manner have been developed.

In a general search system, a string corresponding to a search word may be extracted from a search target material within a classification system, and the search result may be provided by comparing the matched word with the search word. For example, a search system used in an internet bookstore or a library may provide a list of books or summary information according to the Korea Decimal Classification (KDC) classification system. Such a search system may similarly provide various materials similar to a search word in an internet shopping mall or an academic information providing site.

However, in the case of operating the database resources according to various classification schemes, it is difficult to provide a similar material corresponding to the search word by searching the resources by increasing the search target range. For example, in the case of a book search system, not only the classification system of KDC, but also the academic institutions (Korea Research Foundation: KRF: Korea Research Foundation), DDC (Dewey Decimal Classification), UDC (Universal Decimal Classification), etc. The search results according to the classification system of can be hoped. However, the resources according to each classification scheme are interspersed with each other, so it may take a long time to search for resources corresponding to the search terms, and it is not easy to provide a clear search result for each classification scheme. There is a problem with inaccessible information.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and an object of the present invention is to integrate ontologies by integrating database data in different classification systems in various fields in order to provide rich and accurate search results with improved user accessibility. Semantic retrieval method that can recommend classification categories and service meaningful resource search results by mapping them according to the degree of association with classifications in different classification concepts of other fields using classification schemas. To provide a system.

First, to summarize the features of the present invention, the semantic search method according to an aspect of the present invention for achieving the above object, for each classification system divided into two or more classification systems having a plurality of classification categories Databases of resources corresponding to each classification category are generated, and for each keyword, an association relation between the two classification categories is based on similarity values of the first classification category of the first classification scheme and the second classification category of the second classification scheme. And provide a search result by dividing resources belonging to a plurality of classification systems by each classification system or by each classification category, based on the correlation.

The association relationship is to extract each keyword included in any one resource of the first classification scheme and the second classification scheme, the classification of each keyword is the first classification scheme based on the predetermined classification information of each resource. Compute a first similarity value for a category and a second similarity value for a category of the second classification scheme, and calculate the difference between the first similarity value and the second similarity value considering the level of each category. Can be set based on this.

The semantic search method extracts a set of keywords included in each classification category in advance, and matches each keyword with keywords extracted from metadata about a resource including a document file, database information, text data, an image, or a video. A semantic search may be provided by setting and associating associations of classification categories.

The predetermined classification information of each resource includes information on a classification category of a corresponding classification scheme included in a resource or corresponding metadata, and the classification category has a hierarchical structure including at least one level or more detailed categories.

The first similarity value or the second similarity value may be calculated based on a product of a frequency in which a keyword appears in a resource of each classification category and a weight previously set for the classification category to which the resource belongs.

If there is no classification category to which the corresponding resource belongs, in order to use only the frequency in which the keyword appears in the resource, an association relationship with another classification scheme may be adopted by setting all the weights set in advance for the classification categories as the same value. Can be.

When providing the search result, the resources of the classification system including the input keyword are preferentially displayed on the display means for each classification category, and the association system is associated with the classification system adjacent to the resources displayed first. The resources of the different classification systems that are set the most can be displayed for each category.

In addition, the semantic search system according to another aspect of the present invention, a database for storing and managing resources corresponding to each classification category for each classification system divided into two or more classification systems having a plurality of classification categories ; Relationship model setting means for setting an association relationship between two classification categories for each keyword based on similarity values of the first classification category of the first classification scheme and the second classification category of the second classification scheme; And a search result providing means for providing a search result by dividing resources belonging to a plurality of classification systems by each classification system or by each classification category, based on the association, when providing a search result for an input keyword. have.

According to the semantic retrieval method and system according to the present invention, it is possible to integrate resources databased into different classification systems in various fields, and to use the ontology classification schema to separate classifications from other classification concepts in other fields. By mapping the database according to the degree of association and efficiently recommending the classification categories and performing the semantic resource search service, the user can easily and conveniently provide rich and accurate search results.

In order to fully understand the present invention, the operational advantages of the present invention, and the objects achieved by the practice of the present invention, reference should be made to the accompanying drawings which illustrate preferred embodiments of the present invention and the contents described in the accompanying drawings.

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

The semantic search system according to an embodiment of the present invention described below may be implemented to operate in hardware, software, or a combination thereof in a computer, PDA, cellular phone, or other dedicated device, and also may include the Internet, a mobile communication network, Through a wireless Internet network, such as a web (WEB) server, a WAP (WAP) server can be operated to provide a rich and accurate search results easily and conveniently to the user.

1 is a diagram illustrating a semantic search system according to an embodiment of the present invention.

Referring to FIG. 1, the semantic search system according to an embodiment of the present invention has a different classification system such as school classification, KDC, DCC, UCC, classification of products or articles of an Internet portal, and classification of books in an internet bookstore. With regard to the information of the database in which the resources are stored and managed, by establishing and establishing an association relationship between classification categories of different classification systems through a predetermined association setting means, when providing a search result for an input keyword, Based on the same association, a search result may be provided by dividing resources belonging to a plurality of classification systems by each classification system or by each classification category.

For example, the academic classification is A000000 Humanities, B000000 Social Sciences, C000000 Natural Sciences, D000000 Engineering, E000000 Medicine, F000000 Agriculture, Ocean, G000000 Arts, Sports, H000000 Complex, etc. The KDC classification includes 000 resources, 100 philosophies, 200 religions, 300 social sciences, 400 pure sciences, 500 technical sciences, 600 arts, 700 languages, 800 literatures, 900 history, etc. It is used to manage the DDC classification, 000 bibliography, collections, collections, 100 philosophy, 200 religions, 300 social sciences, 400 linguistics, 500 natural sciences, 600 technical sciences (medicine, engineering, agriculture, family studies, etc.), 700 arts. It is used to manage and manage resources of books and related fields such as 800 literature, 900 history, geography and people. In addition, the above-described classification system is applied to Internet portals and bookstores to provide a unique classification system suitable for the purpose of site operation, thereby easily providing search results necessary for a user's keyword search.

Resources in each of these areas may take the form of document files, image files, video files, database information, or text data. Depending on the classification scheme, there may be at least one level of subcategories under the subcategories above. For example, predetermined classification information for a classification category having a hierarchical structure divided into a subclass, a subclass, and a subclass is managed in a database corresponding to the resource or metadata of the resource. The metadata referring to the attribute information of the resource includes not only the information about the classification categories of the corresponding classification scheme, but also the resource's location, keywords, types (text, image, video, etc.), author information, rights conditions, Information regarding usage conditions, usage logs, and the like may be included.

In particular, in the present invention, resources in the corresponding classification system of each field may be constructed in an ontology based on RDF / OWL (Resource Description Framework / Web Ontology Language) syntax. Various resources such as documents, images, videos (videos), database information, and text data provide metadata, and these metadata are maps of semantic relationships centered on classification and keyword concepts. It can be connected to the ontology, and can provide more accurate and rich information to the user through an integrated semantic search in accordance with the association setting according to the present invention.

On the other hand, as shown in the dotted line of Figure 1, the semantic search system according to an embodiment of the present invention, from the information of the database in which the resources corresponding to each classification category is stored and managed for each classification scheme through a predetermined association setting means; For each keyword, an association relationship between classification categories having different classification schemes may be set in advance.

As shown in FIG. 2, each classification system has a classification category having a hierarchical structure, and even when all of a plurality of classification systems are enumerated, it can be seen that an association relationship between categories of different classification systems appears. The present invention proposes a system for generating associations of various characteristics through a semantic relation algorithm. In the semantic search system according to an embodiment of the present invention, a set of keywords frequently used for each category is extracted in advance, and for each keyword, a resource including a document file, database information, text data, an image, or a video is extracted. The association relationship between the classification categories matched with the keywords extracted from the metadata is set and combined. For example, as shown in FIG. 2, a keyword set including, for example, a human name included in each classification category may be generated, and the same keyword may be included in various classification categories. Semantic relationships between them are assigned as associations to provide semantic search.

For example, an association relationship between the two categories may be preset according to the degree of association between each keyword to be input in the first classification category of the first classification scheme and the second classification category of the second classification scheme. In each classification system, each classification category is determined by approaching a concept suitable for the relevant field. In the semantic search system according to an embodiment of the present invention, the schema is established by identifying the associations between the classification categories in advance. In this case, it is possible to improve the accessibility of the corresponding search target resource with respect to the input keyword, and it is easy to set up a relation model such as similarity, accuracy, upper and lower concepts for the search result, and greatly contribute to the recommendation service of the classification category.

For example, as described above, when an association relationship between classification categories having different classification schemes is established for each keyword, the semantic search system according to an embodiment of the present invention may provide a predetermined search result providing means (for example, a search). Engine), by using the semantic search to which the above-described correlation is applied to the input keyword, it is possible to provide a search result by dividing resources belonging to a plurality of classification systems by each classification system or by each classification category. For example, if a keyword to be searched is input, the corresponding category of each classification scheme corresponding to the input keyword is searched first, and accordingly, the search result is provided to the display means by dividing the classification scheme by each classification scheme. It is possible to preferentially recommend information about a resource of a classification category having a high resource to provide a search result. In addition, when providing a search result, the resources of the classification system that includes the most input keywords are displayed first for each classification category, and the resources of the other classification systems that have the highest association with the classification system are displayed. It may be displayed for each classification category.

Hereinafter, a method of setting association relations between classification categories in the semantic search system according to an embodiment of the present invention will be described in detail with reference to FIGS. 3 to 5.

In order to facilitate searching for a large number of resources, a resource may basically include predetermined classification information. Such classification information may be classified into a classification category of a hierarchical structure in each classification system as shown in FIG. 1 used in various fields. Information may be included. Such classification information may be included in metadata that is assigned to a resource as one of attribute information of the corresponding resource.

As such, although the classification information is included in the resource or the metadata, different classification concepts are applied to each classification system to determine classification categories, and thus it is not easy to grasp the association between classification categories of other classification systems. For example, in the KDC or DDC classification system used in the bibliography for the classification of books, it is the same classification category that is exactly matched between the corresponding classification categories such as major classification, subclassification, subclassification, and subclassification, In order to determine whether a classification category exists, there is a difficulty in that the user must look at the search results and classify based on his or her knowledge. Accordingly, the present invention is intended to overcome the difficulty of automatically classifying the classification system and automatically providing the resources of the other classification system with the most established association as a search result. In addition, in the present invention, even when there is no classification category to which the resource belongs (e.g., when there is no classification information), the relevant search results are clearly arranged by automatically associating with a classification category of another classification system. Can be presented and presented.

In FIG. 3 or FIG. 4, the resource or corresponding metadata includes classification information including a predetermined classification category (classification 1-1, 1-2... / Classification 2-1, 2-2...). Each classification category (classifications 1-1, 1-2 ... 1-n) of the first classification system is assigned to one or more classification categories (classifications 2-1, 2-2 ... 2-k) of the second classification system. In order to establish an association relationship, first, a predetermined relationship model setting means includes keywords (keywords 1,2..m) required from language information of a resource, for example, information that can be morphologically analyzed, such as a title, a subject, and a content. We can prepare a learning set by extracting. The keywords (keywords 1,2..m) may be automatically extracted in large quantities by being divided into parts of speech, for example, nouns, adverbs, adjectives, and verbs.

Once the keywords (keywords 1,2..m) are extracted, the relationship model setting means determines that the first similarity value S1 that each keyword has for the classification category (eg, classification 1-1) of the first classification system. And a second similarity value S2 for a classification category (eg, classification 2-1) of the second classification system, and calculates a difference between the first similarity value S1 and the second similarity value S2. You can set up an association based on that.

[Equation 1]

S1 = f _{1 -1} * W _{1 -1} + f _{1 -2} * W _{1 -2} + f _{1 -3} * W _{1 -3} + .. + f _{1 -n} * W _{1 -n} ..

[Equation 2]

S2 = f _{2 -1} * W _{2 -1} + f _{2 -2} * W _{2 -2} + f _{2 -3} * W _{2 -3} + .. + f _{2 -n} * W _{2 -n} ..

Each of the first similarity value S1 and the second similarity value S2 is a frequency f _{1 -1} , f _{1 -2} , f _{1 -3} , in which keywords appear in resources of each classification category, as in the above expressions. f _{1 -n} / f _{2 -1} , f _{2 -2} , f _{2 -3} .. f _{2 -n} ) and the preset weights (W _{1 -1} , W _1-2 ) for the classification category to which the resource belongs. , W _{1 -3} .. W _{1 -n} / W _{2 -1} , W _{2 -2} , W _{2 -3} .. W _{2 -n} ). The above weights may be appropriately set in consideration of the level of the corresponding classification category in each classification system.

In this manner, for each keyword (keywords 1,2..m), the classification category of the first classification system (eg, classification 1-1) and the classification category of the second classification system (eg, classification 2- Association relationship of 1) can be set. For example, when the difference between the first similarity value S1 and the second similarity value S2 is less than or equal to the threshold T%, for example, 10%, the classification category of the first classification scheme (eg, Classification 1-1) As there are many matching keywords between keywords of resources and the classification category of the second classification system (eg, classification 2-1) resources, the classification category of the first classification system (eg For example, classification 1-1) and a classification category (eg, classification 2-1) of the second classification system are set as being related. In addition, when the difference between the first similarity value S1 and the second similarity value S2 exceeds a threshold value T%, for example, 10%, a classification category (eg, Classification 1-1) The keywords of the resources do not match most of the keywords of the classification category (eg, classification 2-1) resources of the second classification system, so that the classification categories of the first classification system (eg, classification 1-1) and the classification category of the second classification system (eg, classification 2-1) are determined to be unrelated. Here, the threshold value (T%) for the difference between the first similarity value S1 and the second similarity value S2, which is the basis for establishing the association, is not limited to 10%, and in some cases, more or more than that. It can be set to the following values.

In this manner, an association relationship between the classification category of the first classification scheme and the classification category of the second classification scheme may be set, and through the association model setting means based on the similarity value for each keyword, as shown in FIG. It is possible to automatically establish a set of relations of categories between the classification schemes and to database and manage them. Accordingly, in the semantic search service system according to an exemplary embodiment of the present invention, a database is classified based on an ontology classification scheme in a different classification scheme in various fields, and is classified with a classification in another classification concept in another field. Based on the mapped associations, it is possible to efficiently recommend classification categories, perform semantic resource search services, and provide users with rich and accurate search results easily and conveniently.

In the setting of the association in the same manner as in FIGS. 3 to 5, when there is no classification category to which the corresponding resource belongs, even if there is no metadata including classification information such as, for example, the classification category, the keyword is associated with the resource. Only the frequency of appearance can be applied to [Equation 1] and [Equation 2] to establish an association in the above manner. For example, preset weights for the classification categories to which the resource belongs (W _{1 -1} , W _{1 -2} , W _{1 -3} .. W _{1 -n} / W _{2 -1} , W _{2 -2} , W _{2 -3} .. W _{2- n} ) all with the same value, and how often the keyword appears in the resource (f _{1 -1} , f _{1 -2} , f _{1 -3} .. f _{1 -n} / f _{2 -1} , f _{2 -2} , f _2-3 .. f _{2- n} ) are applied to Equations 1 and 2 so that the first similarity value (S1) and the second similarity value ( S2) can be calculated. Accordingly, when the difference between the first similarity value S1 and the second similarity value S2 is less than or equal to the threshold value T%, for example, 10%, the corresponding resource belonging to the classification category of the first classification scheme may be determined. 2 It may be adopted and set as being related to the classification category of the classification system. In addition, even when a new keyword is extracted and reflected from a corresponding resource that does not have a classification category as described above, the resource including the new keyword may be included to establish an association relationship with another classification system adopted as described above.

The setting of the association between classification categories of other classification systems may be updated to learn in the same manner as the resource is added, and the association between the categories is accurate for each update, thereby further improving search accessibility. Accordingly, in the search result providing means for displaying the resources of the classification system including the input keyword for each classification category on the display means, the resources of the classification system with high agreement with the input keywords are preferentially for each classification category. In this case, it is also possible to display and recommend resources of another classification system in which the above-described association is most frequently established for each classification category adjacent to the resources indicated above.

As described above, optimal embodiments have been disclosed in the drawings and the specification. Although specific terms have been used herein, they are used only for the purpose of describing the present invention and are not intended to limit the scope of the invention as defined in the claims or the claims. Therefore, those skilled in the art will understand that various modifications and equivalent other embodiments are possible from this. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

1 is a diagram illustrating a semantic search system according to an embodiment of the present invention.

2 is a view for explaining the configuration of the relationship between the hierarchical structure and the majority classification between classification systems.

FIG. 3 is a diagram for describing an association mapping method between a first classification scheme and keywords.

4 is a diagram for describing a method for mapping association between a second classification scheme and keywords.

FIG. 5 is a diagram for describing an association mapping result between keywords and classification categories.

Claims (8)

By dividing the classification system having a plurality of classification categories into two or more categories, the resources corresponding to each classification category are databased for each classified classification system. For each keyword, an association relationship between the two classification categories is set based on the similarity values of the first classification category of the first classification scheme and the second classification category of the second classification scheme. When providing a search result for an input keyword, a semantic search method for providing a search result by dividing resources belonging to a plurality of classification systems by each classification system or by each classification category based on the correlation. The method of claim 1, wherein the association relationship, Extract each keyword included in a resource of either the first classification scheme or the second classification scheme, Calculating a first similarity value that each keyword has to a classification category of the first classification scheme and a second similarity value to the classification category of the second classification scheme based on predetermined classification information of each resource, And a semantic search method based on a difference between the first similarity value and the second similarity value considering the level of each classification category. The method of claim 1, Association of classification categories that match keywords extracted from metadata about resources including document files, database information, text data, images, or videos for each keyword in advance. The semantic search method of providing a semantic search by combining and setting. The method of claim 2, wherein the predetermined classification information of each resource includes information on a classification category of a corresponding classification scheme included in a resource or corresponding metadata. And the classification category has a hierarchical structure including at least one level of detailed categories. The method of claim 2, wherein the first similarity value or the second similarity value, And a keyword is calculated based on a product of a frequency appearing in a resource of each classification category and a weight set in advance for the classification category to which the resource belongs. The classification system of claim 5, wherein in order to use only a frequency in which the keyword appears in the resource when there is no classification category to which the corresponding resource belongs, all other classification schemes are set with the same weights set in advance for the classification categories. And a semantic search method. The method of claim 5, When providing the search result, the resources of the classification system including the input keyword are preferentially displayed on the display means for each classification category, and the association system is associated with the classification system adjacent to the resources displayed first. A semantic search method for displaying resources of different classification schemes which are set the most, for each classification category. A database for storing and managing resources corresponding to each classification category for each classification system classified by dividing the classification system having a plurality of classification categories into two or more categories; Relationship model setting means for setting an association relationship between two classification categories for each keyword based on similarity values of the first classification category of the first classification scheme and the second classification category of the second classification scheme; And Search result providing means for providing a search result by dividing resources belonging to a plurality of classification systems by each classification system or by each classification category, based on the association, when providing a search result for an input keyword. Semantic search system comprising a.

Images