JP2006311482A - User preference extraction apparatus and television program recommendation apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To facilitate user's program selection operation by extracting a television program meeting user's preference and recommending the program to the user. <P>SOLUTION: Programs viewed by a user are sorted into programs viewed periodically everyday or every week and the others and while using information about a program that is viewed although it is not periodically viewed, and information about a program that is meaningful in program contents, other than programs of news or variety shows that the user tends to view because of his/her simple habit, among programs which are periodically viewed, user's preference is extracted and a program meeting the extracted user's preference is recommended. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

The present invention relates to user preference extraction and a television program recommendation method for extracting user preferences and recommending television programs that match the preferences.

With the increase in the number of TV channels, it is difficult for a user to select a favorite program from a TV program guide such as a newspaper or an EPG (electronic program guide). Therefore, a method has been proposed in which user preferences are extracted from the user viewing history, a program that matches the preferences is detected and presented to the user. For example, information on programs that the user has watched for a certain period of time or more is extracted from the EPG, keywords included in the program information are extracted, the user's preference degree for each keyword is obtained from the frequency of keywords, etc., and the user's preference level is strong This is a method for recommending programs including keywords.

In Patent Document 1, there is a possibility that the user does not pay attention to the program simply by matching the channels only by the program that the user has watched for a certain period of time (the channel is selected and received by the tuner). If all of the viewed programs are subject to user preference extraction, correct user preference extraction cannot be performed. Therefore, programs viewed by the user on a periodic basis, such as daily or weekly, and others A method is proposed in which keywords are extracted from only the programs that are classified and viewed periodically, user preferences are extracted using the keywords, and a list of programs that match the extracted user preferences is presented. .

JP 2004-72486 A

  However, there are many cases where programs that are viewed periodically are viewed as mere habits regardless of the content, such as news and wide shows. Many keywords unrelated to user preferences are included. On the other hand, a program that was viewed despite being a single program or not being viewed periodically is a program that the user has selected and viewed because the user's preference matches the program content. Strongly reflects the user's preference. For this reason, there is a problem that the user's preference cannot be accurately extracted only from the program that is periodically viewed, and the appropriate program recommendation cannot be made to the user.

In the present invention, in order to more accurately extract the user's preference, the program viewed by the user is classified into a program that is viewed periodically or every other weekly and a program that is not viewed periodically. Nevertheless, information on programs that have been viewed and information on programs that are meaningful in program content other than programs that are viewed regularly, such as news and wide shows that tend to be viewed simply by habituality. The most important feature is to extract the user's preference using, and recommend a program that matches the extracted user preference.

Since user preference information can be extracted with high accuracy and program recommendation can be performed based on accurate user preference information, the user can easily select an appropriate program from a large number of programs.

Embodiments of the present invention will be described below with reference to FIGS.
FIG. 1 is a diagram showing the configuration of this embodiment. The program viewing information generating means 103, the periodic viewing determining means 106, the content validity determining means 107, the user preference extracting means 108, the unbroadcast program information generating means 111, the preference suitability determining means 113, and the recommended program presenting means 114 are included. It is equipped with the computer. Program individual viewing information table 104, series program viewing information table 1
05, the user preference information table 109, the unbroadcast information table 112, the processing means 103,
106, 107, 108, 111, 113, 114 is a data storage table that is generated, data value set, and referenced.

The program viewing information generating unit 103 includes electronic data of program information 101 such as a program title, a genre, and a performer included in an EPG received by the television receiver or an EPG for the Internet obtained via a network, a television receiver, Electronic data of the viewing history 102 such as a user viewing program and its viewing time and recorded program, in which an operation history for the recording device is recorded and saved, is input. Program individual viewing information table 104 for retaining information on each program and user viewing status, and series program viewing for retaining user viewing information for series programs broadcast continuously every day or weekly Data setting is performed for the information table 105 with reference to the input information. The series viewing information table 105 classifies the programs registered in the program individual information table 104 for each series, and manages the number of broadcasts and the number of viewings for each series program.

The periodic viewability determining means 106 refers to the series program viewing information table 105, determines that a program that the user is periodically viewing such as every day or every week is a periodic viewing program, and shows information indicating that it is a periodic viewing program. This is set in the program individual viewing information table 104. A program viewed by the user other than the periodic viewing program is set as a content preference program, and information indicating that the program is a content preference program is set in the program individual viewing information table 104.

The content validity determination unit 107 determines that a program in which the user's preference is reflected on the program content, not a mere habit, from among the periodic viewing programs obtained by the periodic viewing determination unit 106 is a content effective program, and the content Information indicating that the program is an effective program is stored in the program individual viewing information table 110. For example, it is possible to make a determination using a genre such that a program belonging to a genre such as news or a wide show is not a content valid program, and the other is a content valid program. By separating the content-effective program from the periodic viewing program, it is possible to prevent the use of the user preference degree used when recommending a program content that does not reflect the user's preference.

The user preference extraction unit 108 refers to the program individual viewing information table 104, and sets the information that the content preference program is set in the periodic viewing determination unit 106 and the content validity determination unit 107 as the content effective program. For programs for which information is set, keywords such as performer names included in the program information are extracted, the strength of the user's preference (preference level) for the extracted keyword is obtained, and the keyword and its preference The degree is set in the user preference information table 109. Profiling user preferences by recording what keywords the user is interested in.

The unbroadcast program information generation means 111 receives as input the electronic data of the unbroadcast program information 110 such as the program title, genre, and performer included in the EPG received by the television receiver or the EPG for the Internet obtained via the network. Then, data is set based on the input information in the unbroadcast program information table 112 for holding information on unbroadcast programs.

The preference suitability determining unit 113 refers to the individual program viewing information table 104, the user preference information table 109, and the unbroadcast program information table 112, and adapts to the user preference for each program stored in the unbroadcast program information table 112. The determination result is stored in the unbroadcast program information table 112.

The recommended program presentation unit 114 refers to the non-broadcast program information table 112 and presents the program set with the determination result that the preference suitability determination unit 113 matches the preference as a recommended program to the user. For example, a list of recommended programs may be displayed on a TV screen, a recommended program portion in an EPG displayed on the screen may be displayed in a different color, or a recording instruction for a recommended program may be transmitted to a recording device. In such a manner, the user is notified by voice or TV screen display before the broadcast time of the recommended program.

FIG. 2 is a diagram for explaining the processing method of the program viewing information generating means (103 in FIG. 1).
Referring to the program information (101 in FIG. 1) to be input, the program individual viewing information table (104 in FIG. 1) is set in process 201, and the series program viewing information table (105 in FIG. 1) is set in process 202. Set up.

FIG. 3 is a diagram for explaining the configuration of the program individual viewing information table (104 in FIG. 1). The table is composed of program structure information (301) consisting of titles, subtitles, contents including genre information, performers, broadcast time obtained from data of program information (101 in FIG. 1), and viewing history (102 in FIG. 1). A viewing time 302 to be obtained, viewing presence / absence data 303, a flag storage area 304 indicating the characteristic information of the program, and a pointer storage area 305 to the keyword table 306 are configured. The program configuration information 301 may include broadcast date and broadcast station information. The presence / absence of viewing may be determined based on, for example, the ratio of the broadcast time 301 and the viewing time 302, or may be determined based on whether or not the viewer has watched for a predetermined time. The keyword table 306 is a table that stores titles, subtitles, contents, and keywords obtained from the performers 301 stored in the program individual viewing information table.

FIG. 4 is a diagram for explaining the configuration of the series program viewing information table (105 in FIG. 1). The table is composed of the number of broadcasts 401 of the program of the same title, the number of times of viewing 402 indicating how many times the program was broadcast, and a pointer 403 to the program list 404. Program list 404
Is composed of a pointer 405 and a list pointer 406 to program data of a program having the same title as a target in the program individual viewing information table 104.

FIG. 5 is a diagram for explaining the processing method of the program individual viewing information table setting (201 in FIG. 2). In step 501, the program information (101 in FIG. 1) is referred to, and the title, subtitle, program content, performer, broadcasting station, and broadcast time (301 in FIG. 3) are stored in the program individual viewing information table (104 in FIG. 3). Set. In step 502, program information (10 in FIG.
1) Information on the broadcast start time and broadcast end time of the program and viewing history (102 in FIG. 1)
The viewing time of the program is obtained with reference to the viewing start time and the viewing end time included in. Step 5
03, the viewing time obtained in step 502, the viewing time and the information on presence / absence obtained from the broadcasting time are set in the program individual viewing information table (104 in FIG. 3) (302 and 3 in FIG. 3).
03). The viewing presence / absence is, for example, that the viewing time is a certain time or more, and that viewing is present when the viewing time in the broadcast time exceeds the reference value, and viewing is not performed when the reference value is not exceeded.

FIG. 6 is a diagram for explaining the processing method of the series program viewing information setting (202 in FIG. 2). In step 601, the series program viewing information table (105 in FIG. 4) is referred to and it is determined whether there is a program list (404 in FIG. 4) in which a program having the same title as the title of the target program is registered. To do. If it exists, step 602 is performed, and if it does not exist, step 603 is performed. In step 602, a pointer to the target program data in the program individual viewing information table is stored in a program list in which a program having the same title as the target program is registered, and a pointer indicating the stored program list (403 in FIG. 4). Is added to the number of broadcasts (401 in FIG. 4) in the data of the series program viewing information table in which is stored, and if the viewing presence / absence data (303 in FIG. 3) of the target program is viewed, the number of viewings ( Each data value is updated by adding 1 to 402) in FIG. In step 603, the data of the target program is added to the series program information table. 1 is stored as the number of broadcasts, 1 is stored as the number of viewing times when the viewing presence / absence data is viewed, 0 is stored when there is no viewing, and a pointer to the target program in the program individual viewing information table is stored in the program list. The pointer to the list is stored in the series program viewing information table.

FIG. 7 is a diagram for explaining the processing method of the periodic viewing determination means (106 in FIG. 1). If the periodic viewing condition is satisfied with reference to the series program viewing information table (105 in FIG. 1), in step 701, the program individual viewing information table for the program registered in the corresponding program list (404 in FIG. 4). The flag (304 in FIG. 3) (104 in FIG. 1) is set to a value indicating a periodic viewing program. As a periodic viewing condition, for example, a frequency condition for whether or not a program is watched more than a certain number of times and a ratio condition for whether or not a program is watched for a certain percentage or more of the number of broadcasts of the program are set. It is determined that either or both of the ratio conditions are satisfied. Judgment can be made in consideration of the broadcast frequency and the number of broadcasts, which differ depending on the broadcast cycle and broadcast length of the program. After a value is set in the flag of the program individual viewing information table for all the periodic viewing programs, viewing presence / absence data (303 in FIG. 3) in each program element of the program individual viewing information table
Indicates the presence of viewing, and if the value indicating the periodic viewing program is not set in the flag, in step 702, the value indicating the content preference program is set in the flag.

FIG. 8 is a diagram for explaining the processing method of the content validity determination means (107 in FIG. 1). Referring to the program individual viewing information table (104 in FIG. 1), if the value of the set flag (304 in FIG. 3) indicates a periodic viewing program, it is determined in step 801 whether the content validity condition is satisfied. . A program that satisfies the content validity condition is defined as a content valid program. For example, the genre of the program is not a genre that tends to be viewed as a mere habit, such as news or a wide show, but a genre that has a strong tendency to reflect the user's preference in the program content, such as drama or variety Suppose that the genre of the program is a content valid genre. For a program determined to satisfy the content validity condition in step 801, that is, a program belonging to the content valid genre in the above example, in step 802, the value of the corresponding program data flag in the program individual viewing information table is set to the content valid. Set to a value that indicates the program.

FIG. 9 is a diagram for explaining the processing method of the user preference extracting means (108 in FIG. 1). Referring to the program individual viewing information table (104 in FIG. 1), the value of the flag (304 in FIG. 3) set by the periodic viewing determining means (FIG. 7) and the content validity determining means (FIG. 8) is the content preference. When the value is a value indicating a program or a content valid program, in step 901, text analysis is performed by morphological analysis on the text information of the title, subtitle, program content, and performer (301 in FIG. 3) set in the table. Do. Morphological analysis is a well-known technique for extracting keywords from text information (such as "Introduction to text mining learned with Excel").
In step 902, proper nouns such as performer names and other nouns are selected as keywords from the text analysis result in step 901, and a keyword table (30 in FIG. 3) is selected.
Set to 6). All keywords set in the keyword table referenced from the program individual viewing information table are referred to, and in step 903, the degree of preference for the keyword is calculated. The keyword preference level is obtained, for example, from the appearance frequency of the keyword or the total viewing time of programs including the keyword. In step 904, the keyword and the preference level of the keyword are set in the user preference information table. The keyword is not extracted from the information of the program that is viewed as a mere habit, but the keyword is extracted only from the program whose program content is meaningful to the user, and the preference level is obtained to obtain a value that appropriately represents the user's preference. It is done.

FIG. 10 is a diagram for explaining the configuration of the user preference information table (109 in FIG. 1). It consists of a keyword 1001 and its preference degree 1002.

FIG. 11 is a diagram for explaining the processing method of the unbroadcast program information generating means (111 in FIG. 1). Referring to the input unbroadcast program information (110 in FIG. 1), in step 1101, the unbroadcast program information table (112 in FIG. 1) is displayed in the title, subtitle, program content, performer, broadcast station, broadcast date, broadcast Set the start time and broadcast end time. In step 1102, the text of title, subtitle, performer, and program content data is analyzed by morphological analysis (
Same as 901 in FIG. 9). In step 1103, a keyword is selected from the text analysis result and set in a keyword table referenced from the unbroadcast program information table.

FIG. 12 is a diagram for explaining the structure of the unbroadcast program information table (112 in FIG. 1). The table shows the broadcast date, broadcast station, title, set with reference to unbroadcast program information (110 in FIG. 1).
Each element 1201 includes a subtitle, program content, performer, broadcast start time, broadcast end time, a flag 1202 for storing recommendation information, and a pointer to a keyword table 1204.

FIG. 13 is a diagram for explaining the processing method of the preference suitability determining means (113 in FIG. 1). For each unbroadcast program in the unbroadcast program information table (112 in FIG. 1), refer to the program individual information table (104 in FIG. 1) and the title (301 in FIG. 3) and the title (301 FIG.
1201) and the flag (304 in FIG. 3) has a value indicating a periodic viewing program or a content valid program, step 1301 is performed, and if not, steps 1302 to 1304 are performed. I do. In step 1301, the target unbroadcast program flag (1202 in FIG. 12) in the unbroadcast program information table is set to a value indicating a recommended program. As a result, when the contents of news, etc. are miscellaneous and do not easily reflect the viewer's preference, and the viewer is considered to have a preference for the news program itself, the recommendation is made by determining the matching of the program names. Furthermore, recommendations are made for the dramas watched weekly based on the matching of the titles. In step 1302, the user preference information table (109 in FIG. 1) is referred to determine the program recommendation level of the target unbroadcast program. The program recommendation level is obtained, for example, by referring to the preference level (1002 in FIG. 10) set in the user preference information table for the keyword preference level set in the corresponding keyword list (1204 in FIG. 12). The sum or maximum value of the preference level of keywords in the list. In step 1303, it is determined whether or not the recommendation degree for the unbroadcast program obtained in step 1302 is equal to or greater than a predetermined reference value. If the determination result is equal to or greater than the reference value, the flag (1202 in FIG. 12) of the unbroadcast program information table is set to a value indicating the recommended program. Since a program including many keywords having a high user preference level or a program including a keyword having the highest preference level is selected, a program that matches the user preference can be recommended. Also, for programs that are viewed periodically, the amount of calculation can be reduced because recommendation determination can be made by referring to titles and flags. Further, by making a recommendation determination from two directions of periodicity and program content, a recommendation that matches the user preference can be performed.

FIG. 14 shows an example of a hardware configuration for carrying out the present invention. The hardware for the present invention includes a CPU 1401, a memory 1403 that is stored when the execution program 1402 is executed, and a disk 1404 that stores the execution program 1402. Each of program viewing information generating means 103, periodic viewing determining means 106, content validity determining means 107, user preference extracting means 108, unbroadcast program information generating means 111, preference suitability determining means 113, and recommended program presenting means 114 in FIG. The means is realized by executing the execution program 1402 by the CPU 1401. Program individual viewing information table 104, series program viewing information table 10
5. The user preference information table 109 and the unbroadcast information table 112 are stored in the disk 1404. The program information 101, viewing history 102, and unbroadcast program information 110 are stored in the disc 14
04 may be obtained from an external device via the input / output interface unit I / F 1405 as necessary. The entire hardware configuration shown in FIG. 14 may be a part of a television receiver or a recording device, or may be a separate device connected to them. In the case of a separate device, the device may be a device managed by a service provider connected to a recording device or the like at a user's home via a network, in addition to a device for an individual user installed at the user's home.

It is a figure explaining the structure of this invention. It is a figure explaining the processing system of a program viewing-and-listening information generation means. It is a figure explaining the structure of a program separate viewing-and-listening information table. It is a figure explaining the structure of a series program viewing information table. It is a figure explaining the processing system of a program separate viewing-and-listening information table setting. It is a figure explaining the processing system of a series program viewing information table setting. It is a figure explaining the processing system of a periodic viewing determination means. It is a figure explaining the processing system of a content validity determination means. It is a figure explaining the processing system of a user preference extraction means. It is a figure explaining the structure of a user preference information table. It is a figure explaining the processing system of an unbroadcast program information generation means. It is a figure explaining the structure of an unbroadcast program information table. It is a figure explaining the processing system of a preference suitability determination means. It is a figure explaining the hardware constitutions of this invention.

Explanation of symbols

101 program information, 102 viewing history, 103 program viewing information generating means, 104 individual program viewing information table, 105 series program viewing information table, 106 periodic viewing judging means, 107 content validity judging means, 108 user preference extracting means, 109 user Preference information table, 110 unbroadcast program information, 111 unbroadcast program information generation means, 112 unbroadcast program information table, 113 preference suitability determination means, 114 recommended program presentation means.

Claims (3)

A user preference extraction device that performs user profiling based on program information and user viewing history information, and includes at least an input / output unit for receiving input of the program information, a processing unit, and a storage unit,
In the processing unit, periodic viewing determination for determining whether or not the program viewed by the user is a periodically viewed program based on the viewing history of the user and the program information, and determination that the user periodically views Content validity determination for determining whether the content of the program information is valid for a periodic viewing program that is a program that has been viewed, and content that has not been determined as a periodic viewing program in the periodic viewing determination for a program that the user has viewed User preference is extracted from the information of the content effective program determined to be valid in the periodic viewing program in the target preference program and the content validity determination, and the user preference information is stored in the storage unit A user preference extraction device characterized by: 2. The content validity determination is based on the program information, and determines whether or not the program viewed by the user is effective for extracting the user's preference using a program genre. User preference extraction device. In response to an input of unbroadcast program information about an unbroadcast program from the input / output unit, in the processing device, based on the stored user preference information and the unbroadcast program information, the unbroadcast program becomes a user preference. A user preference extraction method characterized by performing preference suitability determination for determining whether or not it matches, and recommended program presentation for presenting a recommended program that is a program adapted to the user's preference to the user.

Images