KR20030003396A - Method for Content Recommendation Service using Content Category-based Personal Profile structures - Google Patents

Abstract

PURPOSE: A method for a contents recommendation service using a categorized preference profile of a user is provided to recommend the contents preferable to the user from a recommendation candidate list by configuring the preference profile of each user based on the keyword and preference information of the contents. CONSTITUTION: The user's preference profile is configured and stored in a profile database(210) of a recommendation system(201). The system judges that the user is logged on(202). If the user does not log in, a web screen including a general recommendation item(204). If the user logs in, the system requests the preference profile information of the user to the profile database(210) and the profile database(210) transfers the recommendation candidate list to a buffer manager(203, 220). The system decides the recommendation list considering a position of the user(205) and outputs the recommendation list according to the user's position(206).

Description

Method for Content Recommendation Service using Content Category-based Personal Profile structures}

The present invention relates to a content recommendation service method using a user preference profile for recommending content by sector using a user's preference profile on an Internet website, and a computer-readable recording medium recording a program for realizing the method.

As services through the Internet have become a new business area, new ways of generating new profits by providing higher quality content services to users are being proposed.

In the conventional content recommendation service technology, the corresponding content is limited to one product item such as a book and a record, and the basic recommendation is performed through prediction through group propensity analysis, rather than analyzing individual preference information.

Therefore, the accuracy of recommendation for the personal preference content is not very high, and there is a problem in that there is a lack of technical skills in the recommendation method for various web contents such as multimedia contents other than product items.

In addition, even in the content recommendation method, unnecessary information is delivered by showing the previously recommended content to the user at any point in time regardless of the user's location (for example, the same level as the classification level of the content in the website). There was this.

The present invention has been proposed to solve the above-mentioned problems, and comprises a preference profile of each user based on the keyword and the preference information of the content, and the location on the website of the user that the user may prefer It is an object of the present invention to provide a content recommendation service method using a user preference profile that can be selected and recommended from a recommendation candidate list according to the present invention, and a computer-readable recording medium storing a program for realizing the method.

1 is an exemplary configuration diagram of a content recommendation service system to which the present invention is applied.

2 is a flowchart illustrating a content recommendation service method using a user preference profile according to the present invention;

3 illustrates one embodiment of the logical steps of a website according to the present invention.

4 is a diagram illustrating an embodiment of a user preference profile structure for each level according to the present invention;

5 is a diagram for explaining an exemplary candidate candidate list structure for each level according to the present invention;

6 is a diagram illustrating a method for determining a recommendation list according to a location of a user according to the present invention.

7 is a flowchart illustrating an embodiment of a method for configuring a user preference profile according to the present invention.

8 is an embodiment flow diagram for profile matching in accordance with the present invention.

* Explanation of symbols for the main parts of the drawings

101: user 102: web server

103: Recommended System 104: Profile Database

According to an aspect of the present invention, there is provided a content recommendation service method using a user preference profile in a content recommendation service system, comprising: a first step of constructing a user preference profile; A second step of storing the user preference profile in a profile database of the recommendation system and determining a user's login; A third step of outputting a web screen including general recommendations if the user has not logged in; And a fourth step of obtaining a recommendation candidate list by requesting user preference profile information from a profile database of a recommendation system if the user has logged in, and then outputting a recommendation list by reflecting the user's current location on the web content. Characterized in that it comprises a.

On the other hand, the present invention, the content recommendation service system having a processor, the first function of configuring a user preference profile; A second function of storing the user preference profile in a profile database of the recommendation system and determining a user's login; A third function of outputting a web screen including general recommendations if the user has not logged in; And a fourth function of requesting user preference profile information from a profile database of a recommendation system to obtain a recommendation candidate list when the user logs in, and outputting a recommendation list by reflecting the user's current location on the web content. A computer readable recording medium having recorded thereon a program for realizing the present invention is provided.

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

1 is an exemplary configuration diagram of a content recommendation service system to which the present invention is applied.

In the figure, "101" represents a user, "102" represents a web server, "103" represents a recommendation system, and "104" represents a profile database, respectively.

The user 101 may access the web server 102 through a network to receive a content service. In this case, the network performs a plurality of communication connections to perform data communication with each other. Such a network includes the Internet, a wireless communication network, and the like.

The content recommendation service system configures a preference profile of the user 101 according to the category classification of the content. In the actual service, when the user logs in, the profile database 104 of the recommendation system 103 requests information about the user preference profile (for example, a recommendation candidate list, etc.) in advance. This is to speed up the real-time recommendation, and maintains the index of the recommendation candidate list using a buffer manager. Here, in the state where the user is not logged in, the web server 102 shows web content including general recommendations.

On the other hand, when the user's recommendation candidate list is received from the profile database 104 of the recommendation system 103, the user's 101 preference is determined by determining a recommendation list for providing a recommendation according to the user's position from the recommendation candidate list of the user. Provide content according to the profile.

2 is a flowchart illustrating a content recommendation service method using a user preference profile according to the present invention.

As shown in FIG. 2, in the method of recommending content using a user preference profile according to the present invention, first, a user preference profile is configured (201) and stored in the profile database 210 of the recommendation system.

Next, the user's login is determined (202).

As a result of the determination, if the user is not logged in, a web screen including general recommendation is output, and if the user is logged in, the user preference profile information is requested to the recommendation system profile database 210, and the recommendation system profile database 210 is The recommendation candidate list is transferred to the buffer manager 220, and a recommendation candidate list through the recommendation system is obtained (203).

Next, the recommendation list reflecting the user's location is determined (205), and the recommended content according to the user's location is output (206).

The content recommendation service method according to the present invention will be described in more detail as follows.

In the content recommendation service method, a user's preference profile is configured according to the category classification of the content (201). In the actual service, the user recommends the recommendation system profile database 210 for user preference profile information (for example, a recommendation candidate list, etc.) in advance. This is to increase the real-time recommendation speed according to the present invention, and maintains the index of the recommendation candidate list using a buffer manager for this purpose. Here, when the user is not logged in, the web content including the general recommendation is displayed.

On the other hand, if the user's recommendation candidate list is received from the recommendation system profile database 210 (203), the recommendation list for providing a recommendation according to the user's position is determined from the user's recommendation candidate list (205).

In order to determine the recommendation list according to the location of the user, the level of the website should be considered in logical steps, which will be described in more detail with reference to FIG. 3.

3 is a diagram illustrating an embodiment of a logical step of a website according to the present invention.

In the figure, "301" represents the home of the website, "302" represents the content of level 1, "303" represents the content of level 2, and "30m" represents the content of level m, respectively.

In order to determine the recommendation list according to the user's location, the level of the website is divided into logical M stages as shown in FIG. 3.

The structure of the website can be represented as shown in Figure 3 means that the classification of the content is clearly made, the structure and level can be determined according to the category classification of the content.

Therefore, by correlating the level of the website, that is, the classification structure of the content with the location of the user, it is possible to recommend appropriate content according to the location of the user.

4 is a diagram illustrating an embodiment of a user preference profile structure for each level according to the present invention.

In the figure, "401" represents a user ID, "402" represents a website level, and "403" represents a level-specific set.

FIG. 4 illustrates a structure of the user preference profile reflecting this when the user's preference profile is represented by a pair 403 of <keyword, preference>.

The user can be considered to have a set of <keyword, preference> pairs 403 per level 402 for each category group.

5 is a diagram illustrating an example of a recommendation candidate list structure for each level according to the present invention.

In the figure, "501" represents a user ID (ID), "502" represents a website level, and "503" represents a level-specific set.

5 shows a recommendation candidate list according to the level 502 of the website.

That is, a list of recommended content for each of the top N categories designated according to the recommendation policy is determined based on the classified recommendation candidate lists and output to the user.

FIG. 6 is a diagram illustrating a method of determining a recommendation list according to a location of a user according to the present invention. FIG.

In the method of determining a recommendation list according to the location of the user according to the present invention, as shown in FIG. Items are selected as recommended content 603 at the current location, and a recommendation list composed of pairs of similarity with the recommended content 603 is created.

The output of the recommendation content 603 outputs the determined final recommendation list on the user screen. The recommendation content 603 may be output in various ways according to the definition of the user, such as a part of the user screen or a new window.

7 is a flowchart illustrating a method for configuring a user preference profile according to the present invention.

As shown in FIG. 7, the user preference profile configuration method first receives log data from the server group 710 and performs log data preprocessing (701).

Next, the content preference keyword for each category is filtered to transmit the content ID (ID) and user behavior information to the recommendation system 720 to be stored in the profile database 721 of the recommendation system, and the related profile required for updating the user profile. After learning to transmit to the learning server 730, and stores the learning results in the profile database 721 again.

A user preference profile is configured by a set of user-specific <keyword, preference> pairs stored in the profile database 721 (703).

The user preference profile is configured based on the preprocessed log data information to find the preferred keyword information for each category reflecting the content characteristics of the web site as shown in [Table 1].

C C_1 C_2 C_3 C_4 k w k w k w k w C1 6 0.3 14 0.4 11 0.3 . . 4 0.25 12 0.25 14 0.1 . . 3 0.2 47 0.1 34 0.25 . . 2 0.15 26 0.15 31 0.35 . . 7 0.1 23 0.1 . . . . C2 One 0.4 21 0.1 162 0.4 . . 3 0.1 39 0.2 172 0.2 . . 8 0.2 27 0.3 153 0.25 . . 4 0.3 13 0.4 115 0.15 . . C3 65 0.6 79 0.7 . . . . 68 0.4 75 0.3 . . . . ... ... ... ... ... ... ... ... ...

Here, C1, C2, ... denote higher content categories (classifications), and C_1, C_2, ... denote second categories of the corresponding category. And "k" means a keyword, "w" means a preference.

As described above, since the user's preference information is maintained up to a lower category classification, more appropriate information can be provided according to the user's location at the time of content recommendation.

On the other hand, the user preference profile has a set of <keyword, preference> pairs for each category classification preferred by the user, and each time the user uses the service, the relevant values are updated to create a new user's preference profile.

Since the user profile directly reflects the characteristics of the content preferred by the user, the recommendation system 720 may provide a method of easily finding the content for each category preferred by the user.

Meanwhile, input information and output information required for updating a user's preferred profile are required, and a learning server 730 is required for such a task.

The input information includes a set of <keyword, preference> pairs of an existing preference profile, log information such as a feedback type according to the user's behavior, and characteristic information of corresponding contents related to the user's behavior, and the output information includes the user's preference profile. Is a set of updated <keyword, preference> pairs.

In addition, a web server log and a streaming server log of an action after a user accesses a website are required, and the pre-processed log data is a search path of a content page in a site according to a user ID. It includes a view holding time for the content, whether or not the content page bookmarked and printed, the user's behavior information in the content, and the like.

When obtaining data including the user's feedback information as described above is insufficient only by the log of the server itself, other internet technologies that can obtain such information are used, and a purchase log may be added if necessary according to the service characteristics of the site. Can be.

In the present invention, a description of a method of obtaining such log data is omitted, and it is assumed that the initial user profile is simply created by a method such as user input information or collaborative filtering.

8 is a flowchart illustrating an example of profile matching according to the present invention, illustrating a process of creating a recommendation candidate list through profile matching in a recommendation system.

As shown in FIG. 8, profile matching obtains a user preference profile from the profile database 810 of the recommendation system (801) to match the user preference profile with the content profile (802).

Next, it is determined whether the similarity is greater than the threshold (803).

As a result of the determination, if the similarity is greater than the threshold value, the recommendation candidate list is determined (804), and the recommendation list according to the position of the user is determined (805).

As a result of the determination, if the similarity is smaller than the threshold value, the process is terminated as it is.

[Table 2] shows an example of a profile matching process that finds the user's preferred content for the "C2" category of the user's preference profile.

In the content profile, "CID" represents a registration number of each content, and "ai ~ ak" are attributes representing respective characteristics of the content. However, the actual database implementation and storage do not exactly match Table 2 above.

In the above example, as a result of the matching, the contents "i + 1", the contents "i + 3", etc. which are related to the "C2", "C21" categories of the user "JHP" profile are found. The recommendability of the selected contents is determined based on the similarity with respect to the category, and a recommendation candidate list table is constructed. That is, in an actual implementation, a recommendation candidate list for each content category according to a user's preference is stored in a database table.

The present invention provides a method of configuring a user's preferred profile using preprocessed log data, a method of determining a recommendation candidate list for each user through a batch process such as profile matching, and the like based on the above two. The recommendation list is determined according to the location, and each process of presenting the corresponding contents is presented.

As described above, the method of the present invention may be implemented as a program and stored in a recording medium (CD-ROM, RAM, ROM, floppy disk, hard disk, magneto-optical disk, etc.) in a computer-readable form.

The present invention described above is not limited to the above-described embodiments and the accompanying drawings, and various substitutions, modifications, and changes are possible in the art without departing from the technical spirit of the present invention. It will be apparent to those of ordinary knowledge.

As described above, the present invention provides a content personalization service to users of the Internet website, through which the user can selectively receive information necessary for the user's convenience when accessing the website. The effect is to increase the loyalty of the user.

In addition, when the present invention is applied to an e-commerce site, it can be used for intelligent marketing service support, which can contribute to increased sales through differentiation of services.

Claims (7)

A content recommendation service method using a user preference profile in a content recommendation service system, A first step of configuring a preference profile for each content field of the user; A second step of storing the user preference profile in a profile database of the recommendation system and determining a user's login; A third step of outputting a web screen including general recommendations if the user has not logged in; And As a result of the determination, if the user logs in, requesting user preference profile information from the profile database of the recommendation system to obtain a recommendation candidate list, and then outputting the recommendation list by reflecting the user's current location on the web content. Content recommendation service method comprising a. The method of claim 1, The first step is, A fifth step of preprocessing the log data transferred from the server group; A sixth step of finding content preference keyword information for each category based on the log data; And A seventh step of configuring a user preference profile for each category of the user by using the preference keyword information; Content recommendation service method comprising a. The method of claim 2, The log data, The content recommendation service method according to claim 1, wherein the content page search path in the site according to the user identifier, the view holding time for the content, whether the content page is bookmarked and printed or not, and the behavior information of the user made in the content. The method of claim 1, The fourth step, A fifth step of obtaining a user preference profile from the recommendation system; A sixth step of matching the obtained user preference profile with a content profile; A seventh step of determining similarity and threshold values for categories according to the matching result; And An eighth step of determining a recommendation list according to a user's location by determining a recommendation candidate list if the similarity is greater than a threshold; Content recommendation service method comprising a. The method of claim 4, wherein As a result of the determination, if the similarity is smaller than the threshold, the ninth step is excluded from the recommendation candidate list. Content recommendation service method further comprising. The method according to claim 4 or 5, The determining of the recommendation candidate list in the eighth step may include: Substantially, the content recommendation service method comprising constructing and storing a recommendation candidate list table for each content category according to a user's preference in a database table. In the content recommendation service system having a processor, A first function of constructing a user preference profile; A second function of storing the user preference profile in a profile database of the recommendation system and determining a user's login; A third function of outputting a web screen including general recommendations if the user has not logged in; And A fourth function of requesting user preference profile information from a profile database of a recommendation system to obtain a recommendation candidate list if the user has logged in, and then outputting a recommendation list by reflecting the user's current location on the web content; A computer-readable recording medium having recorded thereon a program for realizing this.