KR101157453B1 - System development apparatus and method for personalized service using contet-based recommendation method - Google Patents

Abstract

The present invention relates to an apparatus and method for developing a personalized service system using a content-based recommendation technique. More particularly, the present invention generalizes attribute information that digital content can have and information necessary to provide a personalized recommendation service. The present invention relates to a device and a method for generating a model and a user preference information model, and developing a recommendation engine and a user preference learning engine for content recommendation suitable for a user's preference.

The apparatus for developing a personalization service system using the content-based recommendation technique of the present invention is a model for determining a schema for a content model and a user preference information model using attribute information on digital content and personalization service, and for determining attribute information required for a recommendation service. A model schema editing module generation unit generating a schema editing module; A model schema generation module generation unit generating a model schema generation module for generating a model table for each recommendation service in a database by using the model schema information and recommendation service attribute information expressed in an XML document; A database connection module generation unit generating a database connection module that performs input / output related to the database; A content recommendation module generator for generating a content recommendation module for recommending content suitable for the user's preference information; A preference information learning module generation unit configured to generate a preference information learning module for analyzing user content usage history and learning user preference information; And a client / server module generator for generating a client / server module for providing a recommendation service to the user.

According to the present invention, a GUI-based convenient user interface can significantly reduce the cost and time required to develop a new digital content recommendation service, and a more efficient recommendation service using a formula editor for recommendation algorithms and preference information learning algorithms. Can be provided. In addition, the service can be developed using generalized attribute information, and the service development information can be easily shared using XML, an Internet standard document.

Digital Content, Content-Based Recommendations, Personalization, User Models

Description

System development apparatus and method for personalized service using contet-based recommendation method}

The present invention relates to an apparatus and method for developing a personalized service system using a content-based recommendation technique. More particularly, the present invention generalizes attribute information that digital content can have and information necessary to provide a personalized recommendation service. The present invention relates to a device and a method for generating a model and a user preference information model, and developing a recommendation engine and a user preference learning engine for content recommendation suitable for a user's preference.

Computing technologies have been developed to better understand human beings, and to make their lives better and more convenient. In recent years, personalized recommendation services that support optimized services for user's needs in accordance with social changes that emphasize personality are being actively researched along with the development of computing technology. In addition, with the rapid growth of Internet technology, personalized recommendation services have been applied successfully by various online companies such as Amazon bookstores, CDNow, an Internet CD shop, and MovieFinder, an Internet movie recommendation site.

However, the existing construction method for personalized recommendation service has a problem of designing and implementing a personalized recommendation system and a user model and a content model for each system whenever a new service is provided. For example, if a company that provides Internet TV, news, and various contents on the Internet can build a personalized recommendation service specialized for each content providing service, design a personalized recommendation system, content model, and user preference model suitable for each content providing service. It takes a lot of time and capital to implement. In addition, the personalized recommendation service designed to depend on a specific content providing service has a problem that it is difficult to share user preference information between recommendation services because it does not have formal design information even though it is a service that recommends content with similar characteristics.

An object of the present invention is to generalize various attribute information that digital content can have and attribute information necessary for providing a personalization service, and to provide a method for more conveniently designing and developing a personalized recommendation service using the same.

It is another object of the present invention to provide a method of increasing the efficiency of recommendation by providing a formula editor for designing and implementing a user preference information learning algorithm and a content recommendation algorithm.

An apparatus for developing a personalization service system using a content-based recommendation technique according to the present invention for achieving the above technical problem is to determine a schema for a content model and a user preference information model using attribute information of a digital content and a personalization service. A model schema editing module generation unit generating a model schema editing module for determining attribute information required for the recommendation service; A model schema generation module generation unit generating a model schema generation module for generating a model table for each recommendation service in a database by using the model schema information and recommendation service attribute information expressed in an XML document; A database connection module generation unit generating a database connection module that performs input / output related to the database; A content recommendation module generator for generating a content recommendation module for recommending content suitable for the user's preference information; A preference information learning module generation unit configured to generate a preference information learning module for analyzing user content usage history and learning user preference information; And a client / server module generator for generating a client / server module for providing a recommendation service to the user.
A method of developing a personalized service system using a content-based recommendation technique according to the present invention for achieving the above technical problem is provided. Setting a period; (b) setting attribute information of a content model and a user preference information model in the personalized recommendation service; (c) designing an algorithm for learning a user's preference information using an equation editor; (d) designing a recommendation algorithm that selects content having high preference by calculating a user's preference for each content; And (e) setting a recommendation list construction method, content information retention period and user history retention period to provide a content recommendation service.

delete

According to the present invention, a GUI-based convenient user interface can significantly reduce the cost and time required to develop a new digital content recommendation service, and a more efficient recommendation service using a formula editor for recommendation algorithms and preference information learning algorithms. Can be provided. In addition, the service can be developed using generalized attribute information, and the service development information can be easily shared using XML, an Internet standard document.

The present invention will now be described in detail with reference to the accompanying drawings.

1 is a structural diagram of a personalized service development system using a content-based recommendation method according to the present invention with reference to the accompanying drawings for each module will be described in detail a preferred embodiment of the present invention.

Prior to this, terms or words used in the specification and claims should not be construed as having a conventional or dictionary meaning, and the inventors will appropriately define the concept of terms in order to best explain their invention in the best way possible. It should be interpreted as meaning and concept corresponding to the technical idea of the present invention based on the principle that it can. Therefore, the embodiments described in the specification and the drawings shown in the drawings are only the most preferred embodiment of the present invention and do not represent all of the technical idea of the present invention, which can be replaced at the time of the present application It should be understood that there may be various equivalents and variations.

2 is a flowchart of a process of designing a personalized recommendation service using a model schema editing module to design a content model and a user model, and using the same.

In the basic recommendation service design step, domain names and recommendation service names of digital contents are set to distinguish various recommendation services. In addition, each service sets a learning cycle for user preference information.

In the content model and user preference information model attribute information designing step, a property name, an attribute type, and an attribute value type that configure the content model and the user preference information model for each recommended service are set. The property name refers to a name that distinguishes a property such as genre and channel, and the property type is a type of value that an attribute can have, and is divided into a single type and a multiple type according to the number of values that can have. In addition, the type of attribute value is divided into general type with unlimited range of values that the attribute can have, and index type with limited range. If the type of a specific attribute value is set as index type, the range of values must be set together. Table 1 shows examples and classifications of attribute types and attribute value types that constitute the content model.

In the user preference information learning algorithm design step, an algorithm for learning preference information of each user is designed using an equation editor provided by the present invention. In the content-based recommendation service, learning user preference information consists of analyzing content usage history of each user and updating existing preference information. At this time, for learning user preference information, the formula for calculating attribute preference through user usage history analysis and the formula for reflecting new preference information obtained through history analysis to existing preference information are edited. Here, the preference for the attribute is expressed by quantifying which attribute value the user prefers for each attribute. 3 is a diagram for designing a user preference information learning algorithm according to the present invention.

The equation editor provided by the present invention is used in the user preference information learning algorithm design step and the content recommendation algorithm design step. 4 is a data type used in the formula editor, FIG. 5 is an arithmetic operator available in the formula editor, FIG. 6 is a function available in the formula editor, and in the formula editor, all functions start with '@'. Numeric, character, attribute, and mathematical expressions can be used.

In the content recommendation algorithm design step, a user's preference is calculated for each digital content to design a recommendation algorithm for selecting content having a high preference. The recommendation algorithm is represented by a formula for calculating a user's preference for content. The function that returns affinity information for an attribute with a single attribute type from the user affinity model is @PREFERENCE, and an attribute with a multiple attribute type is selected from four functions (@MAXPREFERENCE, @MINPREFERENCE, @AVGPREFERENCE, @SUMPREFERENCE). Get the preference.

In the recommendation service option designing step, information required for providing a digital content recommendation service is set, and a recommendation list construction method, a content information retention period, and a user history retention period are set.

Here, the recommendation list construction method sets the candidate content selection method for the preference calculation and the number of content to be recommended. The candidate content selection method for calculating preference by the recommendation algorithm is applied differently according to the characteristics of the content. For example, in the case of TV program recommendation, the candidate content list is composed of the contents of the currently broadcasting program and the upcoming program based on the time of receiving the recommendation service. In the case of movie recommendation, the candidate content is currently being screened. It consists of movie content. Accordingly, candidate content should be selected before calculating user preferences for each content. There are a method of generating candidate content, a method of targeting an entire content, a method of targeting only content that satisfies a specific condition, and a method of configuring only content without a record used by a user.

Here, the content retention period refers to the retention period of digital content information to be used for the recommendation service, and there are methods of maintaining the whole and setting an explicit period.

Here, the history retention period means a period for storing a history of the content used by the user, and is set differently for each service according to the characteristics of digital content or a method of generating candidate content. There are two ways to set the history retention period: to save all user history, to delete the history after learning, and to set an explicit period.

7 is an embodiment in which the personalization service information designed through the above process in the present invention is stored in an XML format which is an Internet standard document.

In the present invention, the model schema generation module receives the XML file generated through the model schema editing module as an input, and the model schema generation module uses the XML parser (using the DOM Parser provided by Java) to input the personalized recommendation service. After extracting the design information, the extracted information is transferred to the database connection module.

8 is an embodiment in which personalized recommendation service design information is used in a system. In the drawing, Data Repository refers to a database, and Model_Description, Attribute_Description, Content, User_Pref, and History refers to tables in the database. The solid line means that the inputted design information is entered as a record in each table of the database, and the dotted line is used to create the table using the inputted design information (the CONTENT table using the attribute information constituting the content model). Generation).

In the present invention, the database connection module is responsible for all input and output related to the database. Therefore, the model schema generation module, the training (update) module, the recommendation engine module, and the server module are directly related to the database connection module. 9 is a temporary example of a logical structure of a database used in a personalization service development system using a content-based recommendation technique in the present invention. Referring to the drawings, five tables MODEL_DESCRIPTION, ATTRIBUTE_DESCRIPTION, ENUM_ITEM_LIST, BASIC_USER_INFORMATION, and MODEL_USER_RELATION are generated only once at the time of recording the design information for the personalized recommendation service in the database, and the information is recorded as a record of each table. The CONTENT, HISTORY, and USER_PREF tables are created for each personalized recommendation service with names named '_' and MODEL_ID of MODEL_DESCRIPTION. For example, the contents, history, and user preference table of the personalized recommendation service having MODEL_ID of '1' are 'CONTENT_1', 'HISTORY_1', and 'USER_PREF_1', respectively.

In the present invention, the content recommendation module and the user preference information learning module use the equation editor of the model schema generation module to apply various recommendation algorithms and learning algorithms to the personalization service. Therefore, the developer can easily implement the algorithm by inputting the formulas for the content recommendation algorithm and the user preference learning algorithm for each property.

In implementing the personalized recommendation service development system, although the overall structure may use various methods, the present invention adopts a server / client structure suitable for integrated management of personalized services. In addition, the server and client adopt a structure that communicates by socket. The server is able to perform functions related to driving the learning (update) module for generating a user model and driving the recommendation engine, inputting contents, and registering a user, and used thread techniques to perform requests from multiple clients. .

The following is an embodiment of designing and implementing a digital TV program recommendation service using a personalization service development system according to the present invention. The data used in this example is TV content and history collected in August 2006 and was used as input to the learning (update) module and recommendation engine. Table 2 below is design information for the DTV program recommendation service.

10 is an embodiment of user preference learning for a specific user in the DTV recommendation program service designed as shown in Table 2. Referring to the drawings, it is possible to check the difference between the attribute name, the existing preference for the attribute value, the new preference updated through learning, the existing preference and the updated preference from the first column.

FIG. 11 is an embodiment of recommending a DTV program for a specific user in the DTV recommendation program service designed as shown in Table 2. FIG. Referring to the drawings, the Recommendation Result shows a list of recommended contents, and the lower part of the Recommendation Result shows preference information for each attribute of the user. In the Recommendation Result, each row represents information about one content, and user preferences for that content are shown in the last column, Preference.

Although the present invention has been shown and described with reference to the preferred embodiments as described above, it is not limited to the above embodiments and those skilled in the art without departing from the spirit of the present invention. Various changes and modifications will be possible.

The following drawings attached to this specification are illustrative of preferred embodiments of the present invention, and together with the detailed description of the invention to serve to further understand the technical spirit of the present invention, the present invention is a matter described in such drawings It should not be construed as limited to.

1 is a structural diagram of a personalized service development system using a content-based recommendation technique according to the present invention;

2 is a flow chart of a user model and personalization service design process using the generalized attribute information in the present invention;

3 is a diagram for one embodiment of designing a user preference information learning algorithm according to the present invention;

4 is a data type used in the equation editor provided by the present invention,

5 is an arithmetic operator usable in the equation editor provided by the present invention,

6 is a function available in the equation editor provided by the present invention,

7 is a diagram for one embodiment of storing personalization service design information in an XML document according to the present invention;

8 is a diagram for one embodiment of using personalization service design information in a system according to the present invention;

9 is one embodiment of a logical structure of a user database in the present invention;

10 is an embodiment of learning preferences for a specific user's DTV program in the present invention;

11 is an embodiment of a recommendation list of a DTV program based on preference information of a specific user in the present invention.

Claims (12)

A model schema editing module generation unit for generating a schema for a content model and a user preference information model using attribute information on digital content and personalization service, and a model schema editing module for determining attribute information required for a recommendation service; A model schema generation module generation unit generating a model schema generation module for generating a model table for each recommendation service in a database by using the model schema information and recommendation service attribute information expressed in an XML document; A database connection module generation unit generating a database connection module that performs input / output related to the database; A content recommendation module generator for generating a content recommendation module for recommending content suitable for the user's preference information; A preference information learning module generation unit configured to generate a preference information learning module for analyzing user content usage history and learning user preference information; And Apparatus for developing a personalized service system using a content-based recommendation technique comprising a; client / server module generation unit for generating a client / server module for providing a recommendation service to the user. The method of claim 1, Apparatus for developing a personalized service system using a content-based recommendation technique characterized in that for using the attribute name, attribute type and attribute value type to set the attribute information constituting the content model and the user preference information model. 3. The method of claim 2, The attribute name refers to a name that distinguishes an attribute of a genre or channel, and the attribute type is divided into a single type and a multiple type according to the number of values that the attribute can have, and the attribute value type is a value of the attribute that the attribute can have. Apparatus for developing a personalized service system using a content-based recommendation technique characterized in that it is divided into a general type and an index type according to a range. The method of claim 1, In order to apply various recommendation algorithms and learning algorithms to the personalization service in the content recommendation module and the user preference information learning module, a formula for the content recommendation algorithm and the user preference learning algorithm is input for each property. An apparatus for developing a personalized service system using a content-based recommendation technique. (a) setting a learning cycle for the domain name of the content for the personalized recommendation service, the recommendation service name, and the user's preference information; (b) setting attribute information of a content model and a user preference information model in the personalized recommendation service; (c) designing an algorithm for learning a user's preference information using an equation editor; (d) designing a recommendation algorithm that selects content having high preference by calculating a user's preference for each content; And (e) setting a recommendation list construction method, content information maintenance period, and user history maintenance period to provide a content recommendation service; and a method of developing a personalization service system using a content-based recommendation technique. Claim 6 was abandoned when the registration fee was paid. The method of claim 5, The property information of step (b) includes a property name, an attribute type, and an attribute value type. Claim 7 was abandoned upon payment of a set-up fee. The method according to claim 6, The attribute name refers to a name that distinguishes an attribute of a genre or a channel, and the attribute type is a type of a value that the attribute may have, and is divided into a single type and a multiple type according to the number of values, and the type of the attribute value is the attribute. A method of developing a personalized service system using a content-based recommendation technique, wherein the branch of the value is divided into a general type having a non-limiting range and an index type having a limited range. The method of claim 5, The user preference information learning of step (c) is a method of developing a personalized service system using a content-based recommendation method, characterized in that for updating the existing preference information by analyzing the content usage history of the user. 9. The method of claim 8, In the user preference information learning, a content-based recommendation technique is characterized by editing a formula for calculating property preferences through analysis of user's content usage history and a formula for reflecting new preference information obtained through the history analysis in existing preference information. Development method of personalization service system using. The method of claim 5, The method of constructing a recommendation list in the step (e) is a method of developing a personalized service system using a content-based recommendation method, characterized in that a candidate content selection method for calculating preferences and a number of recommended content are set. The method of claim 5, The content maintenance period of the step (e) is a method of developing a personalized service system using a content-based recommendation technique, characterized in that the retention period of the content to use for the recommendation service. The method of claim 5, The history maintenance period of step (e) means a period of storing the history of the content used by the user.

Images