JP2010288024A - Moving picture recommendation apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a moving picture recommendation apparatus capable of recommending a moving picture scene suitable for user's taste. <P>SOLUTION: The moving picture recommendation apparatus has: a moving picture data storage 12 that stores moving picture data; a scene-attached text data-acquiring part 21 that acquires scene-attached text data attached to each moving picture scene; a content tag-extracting part 23 that extracts a content tag indicating features of each moving picture scene from the scene-attached text data; a topic tag-extracting part 31 that extracts a topic tag related to each set in which the moving picture scene is classified, based on the content tag; a user profile-producing part 81 that produces a user profile indicating user's taste from a viewing and listening operation of the user; and a matching calculation part 63 that generates and outputs recommendation information that recommends the moving picture scene suitable for the user profile among the moving picture data stored in the moving picture data storage 12, based on the content tag and the topic tag. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a moving picture recommendation device that recommends moving pictures to a user who views moving picture data composed of broadcast programs.

  In recent years, systems and services for recommending programs to users based on program metadata such as an electronic program guide (EPG) have been proposed.

  For example, Patent Document 1 uses an electronic program guide (EPG) in order to make it possible to search for a similar program with high reliability even if the amount of program information is small because the document is short like an electronic program guide (EPG). A program search apparatus is described that assigns a keyword based on the characteristics of the content that has been received and recommends a program using preference information set for the keyword.

JP 2009-43156 A

  In the existing video recommendation technology including the invention described in Patent Document 1 described above, the recommended content is a program unit. For example, the user recommends only a specific scene in a program that the user wants to watch, that is, a video scene. I couldn't.

  Also, with a small amount of information such as an electronic program guide (EPG), it was not possible to recommend only a preferable moving image scene.

  The present invention has been proposed in view of such a situation, and an object of the present invention is to provide a moving picture recommendation apparatus capable of recommending a moving picture scene suitable for a user's preference.

  As means for solving the above-described problem, the moving picture recommendation device according to the present invention acquires scene-attached text data that acquires scene-attached text data attached to each moving picture scene obtained by dividing moving picture data configured in units of broadcast programs. A content tag extracting unit that extracts a content tag indicating the characteristics of each video scene from the scene-attached text data acquired by the scene-attached text data acquiring unit, and a moving image based on the content tag extracted by the content tag extracting unit From the user input information obtained from the viewing operation of the user who views the moving image data stored in the moving image data storage unit and the topic tag extracting unit that classifies the scene and extracts the topic tag related to each classified set, User profile indicating user preferences for content tags and topic tags Based on the content tag and the topic tag, the user profile creation unit for creating the video and the video scene that matches the user profile created by the user profile creation unit from the video data stored in the video data storage unit A recommendation information creating unit that creates and outputs recommendation information to be recommended.

  The present invention provides a moving image scene that matches a user profile indicating a user's preference based on a content tag indicating the characteristics of each moving image scene and a topic tag related to a set of moving image scenes classified based on the content tag. Since recommendation information to be recommended is created and output, a moving image scene suitable for the user's preference can be recommended.

1 is a diagram illustrating an overall configuration of a system to which the present invention is applied. It is a figure which shows the whole structure of the moving image recommendation apparatus with which this invention was applied. It is a flowchart for demonstrating operation | movement of a content tag extraction part. It is a figure for demonstrating the specific example of text data attached to a scene. It is a figure for demonstrating the specific example of a content tag. It is a flowchart for demonstrating operation | movement of a topic tag extraction part. It is a figure for demonstrating the specific example of the network of a similar moving image scene. It is a figure for demonstrating the specific example of a topic tag. It is a flowchart for demonstrating operation | movement of a CGM tag extraction part. It is a figure for demonstrating the data memorize | stored in the content tag memory | storage part and the CGM data memory | storage part. It is a figure for demonstrating the specific example of a CGM tag. It is a flowchart for demonstrating operation | movement of a user community tag extraction part. It is a figure for demonstrating the specific example of a user community tag. It is a flowchart for demonstrating operation | movement of a tag integration part. It is a figure for demonstrating the specific example of a scene tag. It is a flowchart for demonstrating operation | movement of a user profile preparation part. It is a figure for demonstrating the specific example of a scene tag. It is a figure for demonstrating the specific example of a user profile. It is a flowchart for demonstrating operation | movement of a matching calculation part. It is a figure for demonstrating the specific example of the user profile memorize | stored in the personal user profile memory | storage part. It is a figure for demonstrating the specific example of the tag regarding the moving image scene which suits a user preference. It is a flowchart for demonstrating operation | movement of a query analysis part. It is a figure for demonstrating the specific example of the query which shows a user preference.

  A video recommendation device to which the present invention is applied is a device that recommends a video to a user who views video data composed of broadcast programs. For example, the video recommendation device is incorporated in a system configuration as shown in FIG. Used.

  That is, the moving picture recommendation device 1 is connected to a tuner 2 that receives broadcast waves, and is supplied with moving image data demodulated by the tuner 2. The moving picture recommendation device 1 is connected to the scene-attached text data server 3, the CGM data server 4, the integrated user profile server 5, and an electric communication line such as the Internet line.

  The scene-attached text data server 3 divides moving image data configured in broadcast program units into moving image scene units, and manages scene-attached text data attached to each divided moving image scene. The scene-attached text data server 3 divides a broadcast program broadcast on a broadcast wave received by the tuner 2 in units of moving image scenes, and for each moving image scene, extracts text data extracted from subtitle data and audio data analysis, It is managed as scene-attached text data. The scene-attached text data server 3 transmits the scene-attached text data to the moving picture recommendation device 1 periodically, for example, for one week.

  The CGM data server 4 manages text information (hereinafter referred to as CGM data) related to CGM (Consumer Generated Media). The CGM data server 4 periodically transmits CGM data obtained by searching, for example, a social networking service on the Internet, an electronic bulletin board, a blog site, and the like to the video recommendation device 1.

  The integrated user profile server 5 manages an integrated user profile in which viewing information of users who use scene-attached text data managed by the scene-attached text data server 3 is integrated. The integrated user profile server 5 integrates the viewing information of each user periodically transmitted from the plurality of video recommendation devices 1 and transmits the integrated information to each video recommendation device 1.

  Using the information transmitted from these servers, the video recommendation device 1 recommends a video scene suitable for the user's preference from the moving image data supplied from the tuner 2 and displays it to the user. Has a configuration as shown in FIG.

  That is, the moving picture recommendation device 1 includes a moving picture data receiving unit 11 and a moving picture data storage unit 12 as shown in FIG. The moving image data receiving unit 11 receives moving image data supplied from the tuner 2. The moving image data storage unit 12 stores the moving image data received by the moving image data receiving unit 11.

  The moving image recommendation device 1 also includes a scene-attached text data acquisition unit 21, a scene-attached text data storage unit 22, a content tag extraction unit 23, and a content tag storage unit 24. The scene-attached text data acquisition unit 21 acquires scene-attached text data transmitted from the scene-attached text data server 3. The scene attached text data storage unit 22 stores the scene attached text data acquired by the scene attached text data acquisition unit 21. The content tag extraction unit 23 extracts content tags indicating the characteristics of each moving image scene from the scene-attached text data stored in the scene-attached text data storage unit 22. The content tag storage unit 24 stores the content tag extracted by the content tag extraction unit 23.

  The video recommendation device 1 includes a topic tag extraction unit 31 and a topic tag storage unit 32. The topic tag extraction unit 31 relates to each set classified by classifying the moving image scenes based on the content tags stored in the content tag storage unit 24 and CGM tags stored in the CGM tag storage unit 44 described later. Extract topic tags. The topic tag storage unit 32 stores the topic tag extracted by the topic tag extraction unit 31.

  The moving image recommendation device 1 includes a CGM data acquisition unit 41, a CGM data storage unit 42, a CGM tag extraction unit 43, and a CGM tag storage unit 44. The CGM data acquisition unit 41 acquires CGM data transmitted from the CGM data server 4. The CGM data storage unit 42 stores the CGM data acquired by the CGM data acquisition unit 41. The CGM tag extraction unit 43 extracts a CGM tag related to scene-attached text data for each moving image scene from the content tag stored in the content tag storage unit 24 and the CGM data stored in the CGM data storage unit 42. The CGM tag storage unit 44 stores the CGM tag extracted by the CGM tag extraction unit 43.

  In addition, the video recommendation device 1 includes a user profile acquisition unit 51, an integrated user profile storage unit 52, a user community tag extraction unit 53, and a user community tag storage unit 54. The user profile acquisition unit 51 acquires an integrated user profile transmitted from the integrated user profile server 5. The integrated user profile storage unit 52 stores the integrated user profile acquired by the user profile acquisition unit 51. The user community tag extraction unit 53 obtains a user community tag indicating the viewing tendency of the user from the content tag stored in the content tag storage unit 24 and the integrated user profile stored in the integrated user profile storage unit 52. Extract. The user community tag storage unit 54 stores the user community tag extracted by the user community tag extraction unit 53.

  The moving picture recommendation device 1 also includes a tag integration processing unit 61, a scene tag storage unit 62, a matching calculation unit 63, and a recommendation display unit 64. The tag integration processing unit 61 integrates the content tag, the topic tag, the CGM tag, and the user community tag described above. The scene tag storage unit 62 stores each tag integrated by the tag integration processing unit 61 in association with each moving image scene. The matching calculation unit 63 uses the scene tag stored in the scene tag storage unit 62 to create recommendation information that recommends a moving image scene that matches the user's preference. The matching calculation unit 63 is a specific processing unit corresponding to the recommendation information creation unit of the present invention. The recommendation display unit 64 displays the moving image scene indicated by the recommendation information created by the matching calculation unit 63.

  In addition, the video recommendation device 1 includes a viewing operation history input unit 71, a viewing operation history storage unit 72, a query analysis unit 73, and a query storage unit 74. The viewing operation history input unit 71 receives a viewing operation and a viewing history as user input information by the user. The viewing operation history storage unit 72 stores the viewing operation history information input to the viewing operation history input unit 71. The query analysis unit 73 analyzes the viewing operation history information stored in the viewing operation history storage unit 72 and creates a query. The query storage unit 74 stores the query created by the query analysis unit 73.

  In addition, the moving image recommendation device 1 includes a user profile creation unit 81, an individual user profile storage unit 82, and a user profile transmission unit 83. The user profile creation unit 81 creates a user profile indicating the user's preference for the scene tag stored in the scene tag storage unit 62 from the query stored in the query storage unit 74. The personal user profile storage unit 82 stores the user profile created by the user profile creation unit 81. The user profile transmission unit 83 transmits the user profile stored in the personal user profile storage unit 82 to the integrated user profile server 5.

  In the video recommendation device 1 configured as described above, specifically, based on the processing illustrated below, a video scene suitable for the user's preference is recommended using the scene tag stored in the scene tag storage unit 62. By displaying it to the user, it is possible to recommend a moving image scene suitable for the user's preference.

  Next, the operation of the content tag extraction unit 23 will be specifically described with reference to FIG.

  In step S <b> 11, the content tag extraction unit 23 acquires scene-attached text data of each moving image scene from the scene-attached text data storage unit 22. For example, the content tag extraction unit 23, as shown in FIG. 4, includes scene-attached text data 100 including a headline 102 that is subtitle information of a moving image scene indicated by a moving image scene identifier 101 and a memo 103 in which audio information is converted into text. To get.

  In step S12, the content tag extraction unit 23 extracts a content tag by performing morphological analysis on the acquired scene-attached text data. Specifically, the content tag extraction unit 23 extracts a noun phrase included in the scene-attached text data as a content tag by morphological analysis.

  In step S13, the content tag extraction unit 23 calculates the weight of each extracted content tag. Specifically, the content tag extraction unit 23 uses an algorithm such as tf-idf to calculate a weight for each content tag using the appearance frequency and the reverse appearance frequency as evaluation indexes.

  In step S14, the content tag extraction unit 23 performs processing for selecting the extracted content tag and processing for integration. For example, the content tag extraction unit 23 selects a content tag based on the weight calculated in step S13. In addition, the content tag extraction unit 23 performs processing for integrating content tags that are similar to each other.

  In step S15, as shown in FIG. 5, the content tag extraction unit 23 creates a content tag 112 to which the weight of the content tag subjected to the processing in step S14 is added for each moving image scene identifier 111, and the content tag Store in the storage unit 24.

  In this way, the content tag extraction unit 23 extracts the weighted content tag as a basic tag representing the feature of the content obtained by performing natural language processing of the attached text data for each moving image scene. To do.

  Next, the operation of the topic tag extraction unit 31 will be specifically described with reference to FIG.

  In step S <b> 21, the topic tag extraction unit 31 determines whether or not an elapsed time from the previous execution of this processing step has passed a predetermined time Tc. Then, the topic tag extraction unit 31 proceeds to step S22 when the predetermined time Tc has elapsed, and ends this processing step when the predetermined time Tc has not elapsed. Here, the predetermined time Tc may be arbitrarily set. For example, the predetermined time Tc may be synchronized with a transmission cycle in which the scene-attached text data server 3 transmits the scene-attached text data.

  In step S <b> 22, the topic tag extraction unit 31 acquires content tags and CGM tags corresponding to each moving image scene from the content tag storage unit 24 and the CGM tag storage unit 44, respectively.

  In step S23, the topic tag extraction unit 31 determines the similarity between the moving image scenes using the weight of the content tag and the weight of the CGM tag, and determines the similarity of each content tag as shown in FIG. Create a network of similar video scenes based on it.

  In step S24, the topic tag extraction unit 31 performs clustering processing of moving image scenes using the network created in step S23, and for example, a plurality of moving image scenes are collected into three sets of topics A to C as shown in FIG. Classify into:

  In step S25, the topic tag extraction unit 31 creates a label for the topic tag related to each set classified in step S24.

  In step S26, the topic tag extraction unit 31 calculates the weight of each topic tag for which a label has been created in step S25. Specifically, for each classified set, the topic tag extraction unit 31 quantifies the tendency of the content corresponding to the set based on the number of moving image scenes, the centrality of the moving image scene, and the burstiness of the moving image scene. Calculate the weight of the topic tag shown.

  In step S27, the topic tag extraction unit 31 creates a topic tag 202 to which the weight of the topic tag is assigned for each moving image scene identifier 201 and stores it in the topic tag storage unit 32 as shown in FIG. In the specific example illustrated in FIG. 8, weighting is performed such that the moving image scene indicated by the moving image scene identifier 201 is included only in the topic A. However, the present invention is not limited to such an example, and a plurality of sets are included. Weighting may be performed so as to be included.

  In this manner, the topic tag extraction unit 31 extracts weighted topic tags as meta tags representing topic information that can be extracted from a set of moving image scenes.

  Note that, from the viewpoint of recommending a video scene suitable for the user's preference, the topic tag extraction unit 31 may extract the topic tag based on at least the content tag, but as described above, it may also be based on the CGM tag. By performing the extraction process, it is possible to classify a more preferable moving image scene and create a topic tag in consideration of social importance such as emotion and reputation.

  Next, the operation of the CGM tag extraction unit 43 will be specifically described with reference to FIG.

  In step S31, the CGM tag extraction unit 43 determines whether or not an elapsed time since the previous execution of this processing step has passed a predetermined time Tc. The CGM tag extraction unit 43 proceeds to step S32 when the predetermined time Tc has elapsed, and proceeds to step S35 when the predetermined time Tc has not elapsed.

  In step S32, as illustrated in FIG. 10, the CGM tag extraction unit 43 includes a content tag storage unit 24 that stores content tags, and a CGM data storage unit that stores CGM data such as a blog 421 and an electronic bulletin board 422. 42, the content tag and the CGM data are obtained respectively.

  In step S33, the CGM tag extraction unit 43 extracts CGM tags by performing morphological analysis on the CGM data acquired from the CGM data storage unit. Specifically, the CGM tag extraction unit 43 extracts keywords relating to the reputation of the moving image content, such as co-occurrence terms and emotional expressions, from the CGM data as CGM tags.

  In step S34, the CGM tag extraction unit 43 calculates the weight of each extracted CGM tag. Specifically, the CGM extraction unit 43 calculates the weight for each CGM tag using the degree of optimism and pessimism of the keyword indicated by the tag and the number of mentions of the keyword in the extraction source text as an evaluation index.

  In step S <b> 35, the CGM tag extraction unit 43 creates a weighted CGM tag 302 for each moving image scene identifier 301 and stores it in the CGM tag storage unit 44 as shown in FIG. 11.

  In this way, the CGM tag extraction unit 43 extracts the weighted CGM tag as a meta tag that represents an emotion or a reputation related to each moving image scene that can be extracted by mining CGM data.

  Next, the operation of the user community tag extraction unit 53 will be described with reference to FIG.

  In step S <b> 41, the user community tag extraction unit 53 determines whether or not an elapsed time from the previous execution of this processing step has passed a predetermined time Tc. Then, the user community tag extraction unit 53 proceeds to step S42 when the predetermined time Tc has elapsed, and proceeds to step S45 when the predetermined time Tc has not elapsed.

  In step S <b> 42, the user community tag extraction unit 53 acquires a content tag from the content tag storage unit 24 and acquires an integrated user profile from the integrated user profile storage unit 52.

  In step S43, the user community tag extraction unit 53, for each moving image data indicated by the content tag, from the integrated user profile, such as the number of playbacks of the entire user and the number of users selected as favorites, are viewed by the user. Information indicating a trend is extracted as a user community tag.

  In step S44, the user community tag extraction unit 53 calculates the weight of the extracted user community tag. Specifically, the user community tag extraction unit 53 calculates the weight of the user community tag by normalizing the number of playbacks of moving image scenes that are different for each moving image scene, the number of users who have made the moving image scenes favorite, and the like. .

  In step S45, the user community tag extraction unit 53 creates a weighted user community tag 402 for each moving image scene identifier 401, and stores it in the user community tag storage unit 54 as shown in FIG. Remember.

  In this manner, the user community tag extraction unit 53 can extract the weighted user community tag as a meta tag representing popularity and attention that can be extracted from the viewing information of the user.

  Note that the user community tag extraction unit 53 uses the integrated user profile acquired from the integrated user profile server 5 to take into account the viewing information of an unspecified number of users without being limited to the usage status of the video recommendation device 1. However, the present invention is not limited to such a form. That is, when the video recommendation device 1 is used by a plurality of users, the individual user profile storage unit 82 stores user profiles corresponding to the plurality of users. For this reason, the video recommendation device 1 uses the user community tag extraction unit 53 to acquire the user community tag from the user profiles stored in the individual user profile storage unit 82 without acquiring the integrated user profile from the outside. May be extracted.

  Next, the operation of the tag integration processing unit 61 will be described with reference to FIG.

  In step S51, the tag integration processing unit 61 determines whether or not the elapsed time from the previous execution of this processing step has passed a predetermined time Tc. The tag integration processing unit 61 proceeds to step S52 when the predetermined time Tc has elapsed, and proceeds to step S54 when the predetermined time Tc has not elapsed.

  In step S <b> 52, the tag integration processing unit 61 acquires a content tag from the content tag storage unit 24. Further, the tag integration processing unit 61 acquires a topic tag from the topic tag storage unit 32. Further, the tag integration processing unit 61 acquires a CGM tag from the CGM tag storage unit 44. Further, the tag integration processing unit 61 acquires a user community tag from the user community tag storage unit 54.

  In step S53, the tag integration processing unit 61 corrects and normalizes the weight of each tag acquired in step S52.

  In step S54, as shown in FIG. 15, the tag integration processing unit 61 creates a scene tag to which each tag normalized in step S53 is assigned to each moving image scene. Here, the scene tag is information in which the weights of the content tag 402, the topic tag 403, the CGM tag 404, and the user community tag 405 are associated with the moving image scene identifier 401 as shown in FIG. is there.

  Next, the operation of the user profile creation unit 81 will be described with reference to FIG.

  In step S <b> 61, the user profile creation unit 81 acquires a query from the query storage unit 74.

  In step S62, the user profile creation unit 81 proceeds to step S63 when the acquired query is an operation command related to viewing of a moving image scene or an operation command related to evaluation of a moving image scene, and ends this processing step otherwise. To do.

  In step S63, the user profile creation unit 81 creates a user profile or updates a previously created user profile by adaptive feedback processing based on the implicit and explicit user behavior history indicated by the acquired query. . For example, the user profile creation unit 81 uses the weights related to the content tag 502, the topic tag 503, the CGM tag 504, and the user community tag 505 associated with each moving image scene identifier 501 as shown in FIG. As shown in FIG. 18, for each user identifier 511 indicating each user who uses the video recommendation device 1, the user's preference regarding the content tag 512, the topic tag 513, the CGM tag 514, and the user community tag 515 is shown by weight. Create or update a new user profile. Then, the user profile creation unit 81 stores the created or updated user profile in the personal user profile storage unit 82.

  Next, the operation of the matching calculation unit 63 will be described with reference to FIG.

  In step S71, the matching calculation unit 63 acquires a content tag, a topic tag, a CGM tag, and a user community tag from the scene tag storage unit 62. In addition, the matching calculation unit 63 acquires a user profile from the personal user profile storage unit 82. In addition, the matching calculation unit 63 acquires a query from the query storage unit 74.

  In step S72, the matching calculation unit 63 proceeds to step S73 when the acquired query is an operation command for displaying recommendation of a similar scene, and proceeds to step S75 otherwise.

  In step S73, the matching calculation unit 63 calculates a moving image scene similar to the query based on each tag stored in the scene tag storage unit 62, and matches the moving image scene that the user indicated by the query was watching last time. Recommendation information for recommending a moving image scene is created.

  In step S74, the matching calculation unit 63 outputs the recommendation information created in step S73 to the recommendation display unit 64, and causes the recommendation display unit 64 to output a screen for recommending a similar moving image scene.

  In step S75, the matching calculation unit 63 calculates the similarity between each tag stored in the scene tag storage unit 62 and each tag indicated by the user profile, and the moving image data stored in the moving image data storage unit 12 Recommendation information for recommending a moving image scene that matches the user profile is created. Specifically, the matching calculation unit 63 creates recommendation information that recommends a personalized moving image scene having the highest degree of similarity with the weight of each tag indicated in the acquired user profile.

  In step S76, the matching calculation unit 63 outputs the recommendation information created in step S75 to the recommendation display unit 64, and causes the recommendation display unit 64 to output a screen recommended as a personal video scene.

  As described above, the matching calculation unit 63 is stored in the query storage unit 74 or the personal user profile storage unit 82, and the content tag 602, topic tag 603, CGM indicated by the user identifier 601 as shown in FIG. A moving image scene that matches the weight of the tag 604 and the user community tag 605 is searched from the scene tag stored in the scene tag storage unit 62. In other words, the matching calculation unit 63 converts the individual moving image scenes having the weights of the content tag 612, the topic tag 613, the CGM tag 614, and the user community tag 615 indicated by the user identifier 611 as shown in FIG. Alternatively, recommendation information recommended as a similar moving image scene is output.

  Next, the operation of the query analysis unit 73 will be described with reference to FIG.

  In step S81, the query analysis unit 73, based on the information stored in the moving image data storage unit 12 and the viewing operation history storage unit 72, the scene viewed last time, the scene viewing information indicating the viewing history, the playback ranking, and the tag designation. Function selection information indicating an instruction is acquired as an input query.

  In step S82, the query analysis unit 73, based on the input query acquired in step S81, the content tag 702, topic tag 703, CGM tag 704, and user community tag 705 indicated by the user identifier 701 as shown in FIG. A query indicating each user's preference expressed by each weight is created. Then, the query analysis unit 73 stores the created query in the query storage unit 74.

  As described above, the video recommendation device 1 is based on the content tag indicating the characteristics of each video scene and the topic tag indicating the topic information related to the set of video scenes classified based on the content tag. Since recommendation information for recommending a moving image scene that matches the user profile indicating the preference is generated and output, it is possible to recommend a moving image scene that is suitable for the user's preference.

  In addition, the video recommendation device 1 creates and outputs recommendation information that recommends a video scene that matches the user profile based on a CGM tag indicating emotion and reputation, so that the user's preference and the social importance of the video content are included. It is possible to recommend a more preferable moving image scene considering the sex.

  In addition, the video recommendation device 1 creates recommendation information for recommending a video scene that matches the user profile, based on user community tags representing the popularity and attention of a plurality of users who use scene-attached text data. To output a more preferable moving image scene in consideration of the user's preference and the preferences of other users.

  In addition, although the moving image recommendation apparatus 1 mentioned above acquires CGM data and an integrated user profile from the outside, and extracts the CGM tag and the user community tag in the moving image recommendation apparatus 1, these extraction processes A function related to the above may be provided on the external server side. That is, the video recommendation device 1 transmits the content tag stored in the content tag storage unit 24 to the CGM data server 4, acquires the CGM tag extracted on the CGM data server 4 side, and stores it in the CGM tag storage unit 44. You may make it memorize | store. The video recommendation device 1 also transmits the content tag stored in the content tag storage unit 24 to the integrated user profile server 5, acquires the user community tag extracted on the integrated user profile server 5 side, and You may make it memorize | store in the community tag memory | storage part 54. FIG.

  In addition, this invention is not limited only to the above embodiment, Of course, a various change is possible in the range which does not deviate from the summary of this invention.

  DESCRIPTION OF SYMBOLS 1 Movie recommendation apparatus, 2 Tuner, 3 Scene attached text data server, 4 CGM data server, 5 Integrated user profile server, 11 Movie data receiving part, 12 Movie data storage part, 21 Scene attached text data acquisition part, 22 Scene attached text Data storage unit, 23 content tag extraction unit, 24 content tag storage unit, 31 topic tag extraction unit, 32 topic tag storage unit, 41 CGM data acquisition unit, 42 CGM data storage unit, 43 CGM tag extraction unit, 44 CGM tag storage Unit 51 user profile acquisition unit 52 integrated user profile storage unit 53 user community tag extraction unit 54 user community tag storage unit 61 tag integration processing unit 62 scene tag storage unit 63 matching calculation unit 64 recommendation display , 71 viewing operation history input unit, 72 viewing operation history storage unit 73 queries analysis unit, 74 queries the storage unit, 81 user profile creation unit, 82 individual user profile storage unit, 83 user profile transmission unit,

Claims (3)

A moving image data storage unit for storing moving image data configured in units of broadcast programs;
A scene-attached text data acquisition unit that acquires scene-attached text data attached to each movie scene obtained by dividing the broadcast program unit of the movie data;
A content tag extraction unit that extracts a content tag indicating the characteristics of each moving image scene from the scene-attached text data acquired by the scene-attached text data acquisition unit;
A topic tag extraction unit that classifies the moving image scene based on the content tag extracted by the content tag extraction unit, and extracts a topic tag associated with each classified set;
A user profile creation unit that creates a user profile indicating user preferences regarding the content tag and the topic tag from user input information obtained from a viewing operation of a user who views the video data stored in the video data storage unit; ,
Creation of recommendation information for recommending, based on the content tag and the topic tag, a video scene that matches the user profile created by the user profile creation unit from the video data stored in the video data storage unit And a recommendation information creating unit that outputs the recommended information. A CGM data acquisition unit for acquiring information on CGM (Consumer Generated Media);
A CGM tag extracting unit that extracts, for each moving image scene, a CGM tag related to the scene-attached text data from the information acquired by the CGM data acquiring unit;
The user profile creation unit is configured to determine a user preference regarding the content tag, the topic tag, and the CGM tag from user input information obtained from a viewing operation of a user who views the video data stored in the video data storage unit. Create a user profile to show
The recommendation information creation unit selects a video scene that matches the user profile created by the user profile creation unit from the video data stored in the video data storage unit, the content tag, the topic tag, and the CGM. The moving image recommendation device according to claim 1, wherein the recommendation is based on a tag. A user user profile acquisition unit for acquiring a user user profile indicating viewing information of a user who uses the scene-attached text data;
A user community tag extracting unit that extracts a user community tag indicating the viewing tendency of the user for each video scene from the user user profile acquired by the user user profile acquiring unit;
The user profile creation unit is configured to obtain user content related to the content tag, the topic tag, and the user community tag from user input information obtained from a viewing operation of a user who views the video data stored in the video data storage unit. Create a user profile that shows your preferences,
The recommendation information creation unit selects a video scene that matches the user profile created by the user profile creation unit from the video data stored in the video data storage unit, the content tag, the topic tag, and the use The video recommendation device according to claim 1, wherein the recommendation is based on a person community tag.

Images