KR20030075112A - Method and apparatus for program recommendation of digital television receiver - Google Patents

Abstract

PURPOSE: A method and an apparatus for recommending a program of a digital TV are provided to make a list of recommendation by directly referring to preference score values preliminarily generated, thereby securing fast response. CONSTITUTION: A program information collector(1) extracts broadcasting program information to be stored in a program database(DB1) and transmits a control signal to a recommending engine controller(4) to preliminarily calculate a preference score related to changed and added program information. A profile manager(2) stores an inner profile of a user in a profile database(DB2) and reports the recommending engine controller changes of the profile. A recommending engine(3) reads program information from the program database and calculates a preference calculation score of each program based on the profile information for recording the result in a preference score database(DB3). The recommending engine controller controls the recommending engine to recalculate the preference score of the changed program for updating the preference score database. A recommended service interface(5) aligns programs according to the corresponding service by directly referring to the preference score values of the preference score database and provides the result of the alignment to the user when the user requests a recommending service.

Description

METHOD AND APPARATUS FOR PROGRAM RECOMMENDATION OF DIGITAL TELEVISION RECEIVER}

The present invention relates to a technology for creating a recommendation list on the basis of viewing preferences for various programs of a user in a digital TV, and in particular, generating a recommendation list in performing an intelligent service such as a virtual channel, an automatic recording, and a program recommendation list. The present invention relates to a method and apparatus for program recommendation of digital TV, which can guarantee fast response speed by reducing the amount of real-time calculation.

Intelligent services in digital TV are being discussed through the TVA-Forum. Among them, there are virtual channels, automatic recording, and program recommendation lists that recommend programs in accordance with individual propensities for broadcasting programs. These personal recommendation services provide individual preferences for each program broadcast on numerous channels of digital TV. It is a service that selects and broadcasts broadcasting information of a program suitable for the user.

Recently, a technology for transmitting a user's preference information about a TV program to a recommendation server, and the recommendation server creates a recommendation list suitable for the individual based on the input information and transmits the recommendation service to the TV. Also appeared.

In the server-client method of transmitting a recommendation result by using a separate server for a recommendation service, a viewing record (or preference information) of an individual user is managed by a server. This restriction creates a problem of privacy invasion. In addition, the server-client method has difficulty in providing recommendation services in various interactive environments due to information delay on the network.

In general, the program recommendation process for an individual requires a lot of calculation as a process of analyzing a target entity according to a user's tendency, comparing them, and selecting a portion thereof. In addition, a program recommendation service for an individual in a digital TV environment should be provided not only for a large amount of programs broadcast on multiple channels but also repeatedly for each user reference. In particular, in the case of the virtual channel service, when the user changes a part of the virtual channel schedule, it may occur that the entire scheduling needs to be performed again.

The My Channel service, which automatically schedules programs to be watched for a certain period of time, such as a day or a week, also analyzes the entire program for a certain period of time, and maximizes the preference level considering the broadcasting time of each program. You should schedule the program hourly so that you can.

That is, since a lot of processing must be performed to provide a my channel service, a general recommendation system has difficulty in providing a fast real-time recommendation service to a user.

In addition, electronic program guide (EPG) information in a digital TV environment is broadcast in program metadata format or downloaded through the Internet or a dedicated channel, which may change from time to time, so personal recommendation services Consideration of the possibility of information changes should provide recommendations based on the latest information.

As such, in the conventional server-client method, an individual user's viewing record is managed by the server, which may cause a privacy invasion problem, and it is difficult to provide a recommendation service in a variety of interactive environments due to information delay on the network. there was.

In addition, since it has to go through a lot of processing to provide my channel service, there was a difficulty in providing a fast real-time recommendation service to the user.

In addition, since the electronic program guide information may change from time to time, there has been a difficulty in always providing a recommendation result based on the latest information.

Accordingly, an object of the present invention is to provide an intelligent TV program recommendation method and apparatus for creating an intelligent program recommendation list based on the latest program broadcasting information and reducing the amount of information computation at the recommendation time point to enable a real-time recommendation service. In providing.

1 is a block diagram of an apparatus for recommending a digital TV according to the present invention;

2 is a signal flowchart of a program information collection process according to the present invention;

3 is a format diagram of a profile table according to the present invention;

4 is a signal flowchart of a profile update process according to the present invention;

5 is a signal flow diagram for a program recommendation processing procedure in the recommender.

*** Description of the symbols for the main parts of the drawings ***

1: Program information collector 2: Profile manager

3: Recommended engine 4: Recommended engine controller

5: Recommended service interface DB1: Program database

DB2: Profile Database DB3: Preferred Score Database

According to a first aspect of the present invention, a program recommendation system according to the present invention is created based on the latest program broadcast information in a digital TV environment.

According to the second aspect of the present invention, the program recommendation system according to the present invention records and manages information necessary for the preference score database or file system in advance in order to improve the recommendation response speed.

According to the third aspect of the present invention, the recommendation result of the recommendation system according to the present invention uses the latest program information at the recommendation time point.

According to the fourth aspect of the present invention, the recommendation system according to the present invention enables recommendation in which the viewing record information is managed and reflected over time.

According to a fifth aspect of the present invention, the recommendation system according to the present invention performs a recommendation function in a digital TV without the help of a separate recommendation server.

The program recommendation method of the digital TV according to the present invention includes a first process of parsing and storing broadcast program information in a program database, and transmitting a control signal to the recommendation engine controller so that a preference score for the changed and added program information is calculated in advance. ; A second step of storing an internal profile of the user including information directly input by the user in a profile database, and notifying the recommendation engine controller of the fact when the profile is changed; Reading program information from the program database, calculating a preferred prediction score for each program based on the profile information of the user of the profile database, and recording the calculation result in a preference score database; A fourth step of, when the control signal or the notification signal is input, outputting a control signal to the recommendation engine to recalculate the preference score for the changed program and add or update the preference score database; When a user requests a recommendation service, a fifth process of sorting a program according to a corresponding service by directly referring to a preference score value of the preference score database and providing the result to the user is performed.

FIG. 1 is a block diagram showing an embodiment of an apparatus for recommending a program for a digital TV according to the present invention. As shown in FIG. 1, program information is parsed and stored in a program database DB1, and changed and added program information. A program information collector (1) for transmitting a control signal to the recommendation engine controller (4) so that a preference score for the precomputation is calculated; The profile manager 2 stores an internal profile of the user including information directly input by the user in the profile database DB2, and outputs a control signal to notify the recommendation engine controller 4 of the fact that the profile is changed. )Wow; Recommended to read the program information from the program database DB1, calculate a preferred prediction score for each program based on the profile information of the user of the profile database DB2, and record the calculation result in the preference score database DB3. An engine 3; When a control signal is input from the program information collector 1 or the profile manager 2, the recommendation engine 3 controls the recommendation engine 3 to recalculate the preference score for the changed program and to add or update the preference score database DB3. A recommendation engine controller 4 for outputting a signal; When a user requests a recommendation service, the system is configured with a recommendation service interface (5) for directly sorting programs according to the service by referring to the preference score value of the preference score database (DB3) and providing the result to the user. When described in detail with reference to Figures 2 to 5 attached to the operation of the present invention configured as follows.

In order to improve response speed by reducing the real-time performance of program recommendations, the results of the recommendation engine (preferred score for each program) according to the user's program preference are pre-calculated and stored in the database (or file system) When requesting a recommendation service, it reads a preference score from the database and immediately performs a recommendation service. However, since the program information may be updated at any time in the digital TV environment, the program information is changed so as to guarantee the reliability of the pre-calculated preference scores, thereby minimizing real-time performance.

Figure 1 shows a preferred embodiment of the digital TV recommendation system according to the present invention. Of course, the present invention can be applied to a recommendation system of an independent recommendation server as well as a recommendation system in a digital TV.

The broadcast program information may be transmitted together with broadcast data, downloaded through an EPG dedicated channel, or additional additional information may be collected through the Internet. The transmitted program information may be retransmitted whenever the broadcasting schedule is changed or additional program information is added.

The program information collector 1 parses the received broadcast program information and stores it in the program database DB1, manages the stored program information, and sends a control signal to the recommendation engine controller 4 with respect to the changed and added program information. (CS1) is transmitted so that the preference score is calculated in advance.

The information collection process of the program information collector 1 will be described in more detail with reference to FIG. 2 as follows.

When the program information is received, the program metadata information is parsed, and if the received program information is new program information, and the program information is found, the program information is added to the program database DB1 (SA1-SA4).

However, if it is determined that the received program information is not new program information, it is checked whether it is changed information for the existing program, and if it is found to be changed information, the program database DB1 is updated. (SA5, SA6)

If it is determined that the changed program information is information related to the preference score, and the information is found to be the information, it is requested to the recommendation engine controller 4 to record the changed preference score in the program database DB1. (SA7, SA8)

The profile manager 2 stores the information directly input by the user and an internal profile calculated from the user viewing information in the profile database DB2, and also manages the previously stored profile and changes the profile. When the control signal CS2 is transmitted to the recommendation engine controller 4, the fact is informed.

The internal profile is an incremental update that is periodically accumulated viewing statistical data from user viewing information (e.g., additional information such as viewing records, recording records, and input information on whether a program is preferred). do.

As shown in FIG. 3, the internal profile includes cumulative statistical values and period statistical values for each item. Here, the cumulative statistical value is a value used for the actual recommendation, and the period statistical value is a statistical value for the viewing record generated after the last update time and is reflected at the next program update time. The cumulative viewing history value of the profile is updated to reflect the latest viewing history every certain period. At this time, the past cumulative statistics are reflected by the weight of time coefficient as shown in the following [Equation 1], so that the recent viewing history is more Highly reflected.

The update process of the profile will be described in more detail with reference to FIG. 4 as follows.

If it is found that the user's preference information is directly input, the profile database DB2 is updated and the control signal CS2 is output so that the recommendation engine controller 4 recalculates the preference score. (SB1-SB3)

However, if the user is found to be watching any program, the period statistics of the internal profile associated with the watched program are increased (SB4-SB9).

In addition, if it is found that the internal profile is periodically updated, the values of the entire internal profile are read, and the cumulative statistical values of all items are recalculated using time coefficients and period statistical values, and the recommendation engine controller 4 is preferred. The control signal CS2 is output to recalculate the score (SB10-SB17).

Meanwhile, the program recommender includes a recommendation engine 3 and the recommendation engine controller 4. The recommendation engine 3 reads the program information from the program database DB1, calculates a preferred prediction score for each program based on the profile information of the user of the profile database DB2, and calculates the result of the preference score database. Write to (DB3). The preferred prediction score is calculated as a value that can be absolute compared in consideration of a broadcast time, a channel, a genre, a character, a keyword, and the like of the program. Therefore, since the preference prediction score has the same score value regardless of the point of time when the user's profile is not changed, the preferred prediction score can be generated and managed in advance.

The recommendation engine controller 4 is a module for calculating and storing a preference score value in advance. When the program information or profile information is changed, the recommendation engine controller 4 manages the preference score value of the preference score database DB3. That is, when the control signals CS1 and CS2 are input from the program information collector 1 or the profile manager 2, the recommendation engine 3 recalculates the preferred scores for the changed programs, and generates a preference score database. The control signal CS3 is output to add or update to DB3.

The recommended engine control process of the recommended engine controller 4 will be described in more detail with reference to FIG. 5 as follows.

When new program information is transmitted from the broadcasting station and the recommendation engine controller 4 receives the control signal CS1 from the program information collector 1, the recommendation engine controller 4 outputs the control signal CS3 to the recommendation engine 3, As a result, the recommendation engine 3 generates a preference score value and adds the value to the preference score database DB3. (SC1-SC4)

When the update request is generated due to the change of program information, the preference score value is recalculated to update the preference score database DB3 (SC5-SC8).

When the profile information is changed, that is, when the recommendation engine controller 4 receives the control signal CS2 from the profile manager 2, all the information of the target program is read out to recalculate the preferred score, and thus the preferred score database DB3. Update all the values of (SC9-SC14).

On the other hand, when the user requests the recommendation service, the recommendation service interface 5 immediately refers to the preference score value of the preference score database DB3, sorts the program according to the corresponding service, and provides the result to the user.

As a result, in the program recommendation system according to the present invention, when a user requests a program recommendation service, the user directly requests the program recommendation service by directly referring to a previously generated preference score value without generating a preference score by the recommendation engine 3. The program recommendation is possible only by sorting the program.

As described in detail above, in the present invention, when a user requests a recommendation service for a broadcast program, a recommendation list is created by directly referring to a previously generated preference score value without generating a preference score by the recommendation engine, thereby reducing the amount of real-time computation. It is effective to guarantee fast response speed.

Therefore, in addition to enabling real-time recommendation service in a limited processor such as a digital TV set-top box, it also significantly reduces real-time computation performance even when the recommendation engine operates on an independent recommendation server. An excellent effect is expected to provide.

In addition, by generating and updating a profile reflecting the viewing history information of the user over time, it is possible to flexibly respond to the pattern change of the user's program.

Claims (9)

A first step of storing and managing broadcast program information using a program database, and transmitting a control signal to the recommendation engine controller so that a preference score for the changed and added program information is calculated in advance; Storing a user's internal profile including information directly input by the user in a profile database, and notifying the recommendation engine controller of the fact when the profile is changed; Reading program information from the program database, calculating a preferred prediction score for each program based on the profile information of the user of the profile database, and recording the result in a preference score database; A fourth step of, when the control signal or the notification signal is input, outputting a control signal to the recommendation engine to recalculate the preference score for the changed program and add or update the preference score database; When a user requests a recommendation service, a method of recommending a program for a digital TV, comprising a fifth process of immediately sorting a program according to a corresponding service by referring to a preferred score value of the preferred score database and providing the result to the user. . The method of claim 1, further comprising: a first step of confirming that new program information is received and adding the new program information to a program database; A second step of updating the program database if the existing program is found to be changed information; And a third step of requesting a recommendation engine controller to record the changed preference score in the program database if the changed program information is found to be information related to a preference score. The method of claim 1, wherein the internal profile in the second process is periodically updated and updated as viewing statistics data accumulated over time from viewing information such as viewing information, recording information, and additional information such as input information on whether a program is preferred. Program recommendation method of the digital TV, characterized in that. The method of claim 1, wherein the internal profile of the second process is a value used for the actual recommendation, the cumulative statistical value for each item that is updated to a value reflecting the latest viewing record at a certain period, and the viewing record generated after the last update time The program recommendation method of the digital TV, characterized in that consisting of a statistical value for the period statistics reflected at the next program update time. 5. The method of claim 4, wherein the new cumulative statistical value for the cumulative statistical value is calculated by the following Equation. The method of claim 1, wherein the internal profile of the second process comprises: a first step of updating a profile database and recalculating a preference score in the recommendation engine controller when the user's preference information is directly input; If the user watches any program, increasing a period statistical value of an internal profile associated with the watched program; When the internal profile is periodically updated, the cumulative statistical values of all items are read and recalculated using the time coefficient and the period statistical value, and the recommendation engine controller is updated to the third step of calculating the preferred score. The program recommendation method of the digital TV, characterized in that the processing. The method of claim 1, wherein the preferred prediction score of the third process is calculated as a value that can be compared in absolute terms in consideration of a broadcast time, a channel, a genre, a character, a keyword, and the like of the program. The method of claim 1, further comprising: a first step of generating a preferred score value with a recommendation engine and adding it to the preferred score database when new program information is input; A second step of updating the preference score database by recalculating the preference score value when program information is changed and an update request is generated; When the profile information is changed, the program recommendation method of the digital TV, characterized in that the third step of reading all the information of the target program to recalculate the preference score to update all the values of the preference score database. A program information collector (1) for extracting broadcast program information and storing it in a program database (DB1) and transmitting a control signal to the recommendation engine controller (4) so that a preference score for the changed and added program information is calculated in advance; A profile manager 2 for storing an internal profile of the user including information directly input by the user in the profile database DB2 and notifying the recommendation engine controller 4 of the fact when the profile is changed; Recommended to read the program information from the program database DB1, calculate a preferred prediction score for each program based on the profile information of the user of the profile database DB2, and record the calculation result in the preference score database DB3. An engine 3; When the control signal is input or the profile change notification signal is input, the recommendation engine 3 causes the recommendation engine 3 to recalculate the preference score for the changed program and add or update the control signal to the preference score database DB3. An engine controller 4; When a user requests a recommendation service, the user can refer to the preference score value of the preferred score database (DB3) to sort programs according to the corresponding service, and then configure the recommendation service interface (5) for providing the result to the user. Digital TV program recommendation device.