JP2012070175A - Content recommendation apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a content recommendation apparatus that can present content matching a user's preferences.SOLUTION: The content recommendation apparatus includes: a first acquisition section for acquiring first meta information included in first content viewed by a user and the user's view history; a second acquisition section for acquiring second meta information included in second content to which structured information is added; a calculation section for calculating a feature quantity set using scores of feature quantities acquired from the view history and correlations between the feature quantities; a determination section for determining whether the second content is related with the first content included in the view history according to the feature quantity set; and a display section for displaying the second content in accordance with the determination result of the determination section.

Description

  Embodiments described herein relate generally to a content recommendation device.

  Patent Document 1 discloses a technique for presenting a relationship between a moving image in broadband broadcast via the Internet and information on a broadcast program such as a television broadcast.

Japanese Patent Laid-Open No. 2008-211274

  The content obtained via the Internet is different from the content such as a program broadcast on a television or a video-on-demand, and the information added to the content is not structured. Therefore, when a technique such as word matching is simply used, there is a problem that appropriate content for the user is not recommended.

  The problem to be solved by the present invention is to provide a content recommendation device capable of presenting content that meets user preferences.

  In order to solve the above problem, a content recommendation device according to an embodiment of the present invention obtains first meta information included in first content viewed by a user and the viewing history of the user. A second acquisition unit that acquires second meta information included in the second content to which the structured information is added, a score of the feature amount obtained from the viewing history, and the feature amount A calculation unit that calculates a feature amount set using correlation, and a determination unit that determines whether or not there is a relationship between the first content and the second content included in the viewing history using the feature amount set; And a display unit for displaying the second content using a determination result of the determination unit.

The figure showing the structure of the content recommendation apparatus which concerns on this embodiment. The figure which shows an example of the content recommendation apparatus which concerns on this embodiment. The flowchart which shows the process of the content recommendation apparatus which concerns on this embodiment. The figure which shows an example of a content search. The flowchart which shows the example of the process in the calculation part of this embodiment. The flowchart which shows the example of a feature-value list production | generation process. The flowchart which shows the example of feature-value evaluation information calculation. The flowchart which shows the example of feature-value set production | generation. The figure which shows the example of a feature-value set. The flowchart which shows the example of the process in the determination part of this embodiment. The flowchart which shows the example of the process in the update part of this embodiment. The figure which shows the example of a display in the display part of this embodiment.

  With the digitization of content and the spread of the Internet environment, video that can be used on the Internet has increased dramatically. Since the content and timing of viewing the video on the Internet can be freely selected, the user can enjoy content according to his / her interests and preferences. For example, in a net video distribution service in which general users can post videos and other users can freely view these videos, a huge number of videos are constantly posted and viewed. Such a net moving image distribution service generally provides a moving image search function, and a user can find a moving image that the user likes by searching for a moving image with a keyword or the like. In addition, since a moving image distributed by the Internet moving image distribution service often has a length of one moving image from several seconds to several minutes, the user tends to easily and repeatedly watch a favorite moving image. From the above characteristics, it is considered that the keywords used by the user for searching and the content to be viewed in the Internet video distribution service strongly express the user's preference.

  On the other hand, unlike the contents of broadcast programs and video-on-demand services, online videos are not structured, and words of various characteristics such as titles, tags, names of people, places, and song titles. Are often mixed. For example, the movie “Sky”, which is sung by singer A, will be described as an example. When a content including “Singer A” is presented to a user who has watched a video whose title is “Singer A, Sky”, there is a high possibility that content matching the user's interest can be presented. On the other hand, content including only the word “sky” does not necessarily match the user's interest. This means that when a broadcast program is recommended by a technique such as word matching, an appropriate broadcast program is not necessarily recommended.

  In the present embodiment, the meta information obtained from the viewing history of the Internet video content is structured as a feature amount set including performers and keywords representing the user's preference, and content such as a broadcast program is represented using each feature amount set. Recommendation to. This assists user viewing.

  Embodiments for carrying out the invention will be described below.

(First embodiment)
FIG. 1 is a diagram showing a configuration of a content recommendation device 10 according to an embodiment of the present invention. In the present embodiment, content that can be viewed as a first content on a video sharing service or the like (hereinafter referred to as “net video”), and content that can be viewed as a second content on a television or the like (hereinafter “ This will be described as an example.

  The content recommendation device 10 according to the present embodiment includes a first acquisition unit 101 that acquires meta information obtained from a history of a viewed online video when a user views the online video, program information associated with a broadcast program, and the like. A second acquisition unit 102 for acquiring the meta information, a calculation unit 103 for calculating a feature amount set from the meta information obtained by the first acquisition unit, a broadcast program and a net video from the obtained feature amount set A determination unit 104 that determines whether or not there is an association, an update unit 105 that updates a viewing history of an online video, a display unit 106 that displays a recommended broadcast program, and a first storage unit 107 that stores a feature amount set The second storage unit 108 stores the user's viewing history.

  FIG. 2 is a diagram showing an example of the overall configuration including a network such as the Internet and television broadcasting. As shown in the figure, the content recommendation device 10 acquires the viewing history stored in the net video viewing history storage unit 12 via the net video viewing control unit 11 from the net video distribution server. Information associated with the broadcast program stored in the broadcast program information storage unit 22 via the broadcast program viewing control unit 21 from the broadcast server that distributes the broadcast program, for example, the program title, start time, end time, program Get information such as information about the contents of. The net moving image viewing history storage unit 12 includes a function of the second storage unit 108 that stores the moving image viewing history of the user. In response to a request from the content recommendation processing device 10, the net moving image viewing history storage unit 12 stores the viewing history related to the user's net moving image, and updates the viewing history related to the net moving image by new search or viewing.

  The net moving image viewing control unit 11 is connected to a net moving image distribution server that distributes net moving images over the Internet, and receives keywords used for content search from the display unit 106 and the update unit 105. The received search keyword is transmitted to the Internet video distribution server on the Internet by the network video viewing control unit 11, and the search list received from the server is transmitted to the display unit 106 in a predetermined format. The net video viewing control unit receives information about the viewing status of the user's net video from the search information acquisition unit and display unit 106 that stores the search list in a predetermined format in the net video viewing history storage unit 12, and stores the net video viewing history A viewing state update unit that updates the viewing information of the content in the department, a video identification number (video ID) included in the content or information (URL information, etc.) regarding the video storage location in the server from the display unit 106, You may have a moving image reproduction part which transmits information to a net moving image delivery site, encodes the moving image information delivered from a server, and transmits to the display part 106 in the form which a user can view.

  The broadcast program viewing control unit 21 is connected to a broadcast server that receives broadcast waves and distributes the broadcast wave to the terminal, and stores broadcast program information from the broadcast server in the broadcast program information storage unit 22 in a predetermined format. An acquisition unit, a viewing information transmission unit that outputs information stored in the broadcast program information storage unit to the display control unit 35, video information received from the display control unit, and encoded video information distributed from the server And a broadcast program playback unit that transmits to the display unit 106 in a form that can be viewed.

Table 1 shows an example of broadcast program information stored in the broadcast program information storage unit 22. Broadcast program information is configured based on general electronic program guide information transmitted from broadcast stations such as program ID, title, performer, genre, and keyword.

  Here, the operation in the content recommendation device 10 will be described with reference to FIG. When the user views a net video, the content recommendation device 10 sends the net video viewing history acquired by the first acquisition unit 101 to the calculation unit 31, and sets a feature amount representing the user's preference from the net video viewing history. Is generated (step S31). The generated feature quantity set is stored in the first storage unit 107 (step S32).

  The generated feature amount set is sent to the determination unit 32. The determining unit calculates the degree of association between the feature amount and the broadcast program information from the obtained feature amount set, and determines whether or not the feature amount set is associated with the broadcast program information (step S33). When there is a relationship, a related broadcast program identification number (hereinafter referred to as a program ID), a feature amount ID for identifying the generated feature amount set, and a feature amount relevance degree are created as a related program list. The created related program list is sent to the display unit 106 and displayed as related content related to the net moving image (step S35).

  The stored feature value set may be generated by the update unit 34 as a search term set, and each element of the search term set may be transmitted to the search information acquisition unit of the network video viewing control unit 11. By generating a search term set, the search history of content that the user is interested in is automatically updated (step S34). New related content may be displayed on the display unit 106 using the updated history (S35).

  Further, in the case of an apparatus capable of viewing both a net moving image and a broadcast program, the display unit 106 may include a viewing content selection unit that selects whether to view the net moving image or the broadcast program. In general, information included in a broadcast program has a regular arrangement of information such as words representing the title and contents of the broadcast program, names of performers, and the like. For broadcast programs having information structured in a regular arrangement, the information contained in the net moving image has a small amount of information itself and no regularity. For this reason, since the handling and viewing method of the net moving image and the broadcast program are greatly different, it is preferable that the net moving image and the broadcast program can be individually operated.

  Similarly, the display unit 106 may include a net video search display unit that transmits a search term input by the user to the search information acquisition unit of the net video viewing control unit 11 and displays a search result to the user. In addition, a video playback request unit that transmits the video ID of the video selected by the user from the search results to the video playback unit of the network video viewing control unit 11 and receives the video stream, and information on the viewing state when the video viewing ends. The user selected from the viewing state transmission unit that transmits to the viewing state update unit of the net video viewing control unit 11, the broadcast program information display unit that displays the broadcast program information from the broadcast program viewing control unit 21, or the broadcast program information You may have the program reproduction request part which transmits program ID of a broadcast program to the broadcast program viewing control part 21, and receives a broadcast stream.

  FIG. 4 is a diagram showing an example of a search screen in the case where a net moving image search is performed on the display unit 106. A search word is input in the net moving image search display unit, and a moving image is reproduced from the search result. Here, the user performs a search by inputting a search word into the search word input unit.

In this embodiment, it is assumed that the user performs a search using the search term “singer A”. An example of the search list at this time is shown in Table 2. The search list includes a net moving image including singer A as a search word, and in addition to the moving image title, information on tags and URLs attached to the moving image separately from the title can be obtained. The information in Table 2 is presented to the user through the display 106 by the search information acquisition unit of the net video viewing control unit 11 and simultaneously stored in the net video viewing history storage unit 12.

Table 3 shows an example of the net video viewing history stored in the net video viewing history storage unit 12. The net moving image viewing history is composed of search list information, words used for the search, and viewing state information. For example, since the net moving image of the title “Singer A_Sky_PV” (“_” indicates a blank) is included in the search list by the search word “Singer A”, “Singer A” is given. The search term information may be performed with a plurality of words. The viewing state represents a state where the user views each net moving image. When the user performs a search, the result is stored in the net moving image viewing history storage unit 12 as the viewing state of all the net moving images is “unviewed”.

  When the user selects a video to be viewed from the search list, the video ID of the video selected by the video playback request unit of the display unit 106 is transmitted to the video playback unit and the viewing state update unit of the net video viewing control unit 11. Thereby, the viewing state of the net moving image viewing history can be updated. In Table 3, the user watched the net video of the title “Singer A_Sky_PV” to the end (displayed as “complete”), and watched the net video of “Singer A_Sky_live” halfway (“Suspend”). Example).

  Next, individual operations of the calculation unit 103, the determination unit 104, and the update unit 105 constituting the content recommendation device will be described with reference to the drawings.

  FIG. 5 is a flowchart illustrating a specific example of processing in the calculation unit 103.

  The calculation unit 103 generates a feature amount list including the feature amounts extracted from the online video viewing history (S311). For each feature quantity in the generated feature quantity list, feature quantity evaluation information including a feature quantity score for representing the user's preference and a feature quantity correlation matrix for representing a correlation between the feature quantities is calculated (S312). . One or more feature quantity sets representing the user's preference are generated using the calculated feature quantity evaluation information (S313). The calculated feature value set is stored in the first storage unit 107 (S314).

  FIG. 6 is a flowchart showing a specific example of the feature quantity generation processing in S311.

  First, the calculation unit 103 generates an empty feature amount list having no elements (S3110). Next, it is determined whether or not there is an unselected net moving image in the net moving image viewing history (S3111). If all the net moving images have not been selected, the process proceeds to S3112. . Next, the calculation unit 103 selects a net video that has not yet been selected from the net video viewing history (S3112). It is determined whether or not there is a feature quantity not selected among the feature quantities of the selected net moving image (S3113). If all the net moving images have not been selected, the process proceeds to S3114. If selected, the process proceeds to S3111. move on. In this embodiment, the word used for the search and the tag information attached to the net moving image are used as the feature amount of the net moving image. Here, the tag information is composed of words obtained by disassembling the title by morphological analysis and words given by the poster when the moving image is posted. In the example of Table 3, the feature quantity of “Singer A Sky PV” is “Singer A” used for the search, and “Sky”, “PV”, “New song”, and “Cool” in the tag information.

  The calculation unit 103 selects a feature amount that has not yet been selected from the feature amounts of the selected net moving image (S3114). It is determined whether or not the selected feature amount exists in the feature amount list (S3115). If the selected feature amount does not exist in the feature amount list, the process proceeds to S3116, and the selected feature amount is already a feature amount. If it exists in the list, the process proceeds to S3113. If not present in the feature quantity list, the selected feature quantity is added to the feature quantity list (S3116).

Table 4 is an example of the feature amount list generated by the calculation unit 103 in the process of S311. The feature amount list includes a feature amount ID for identifying a feature amount, a feature amount content representing the content of the feature amount, and a feature amount attribute representing an attribute of the feature amount.

  FIG. 7 is a flowchart showing a specific example of the feature amount evaluation information calculation processing in S312.

The calculation unit 103 generates an initial feature value score vector for storing the score of the feature value list (S3120). An example of the initial feature amount score vector is shown in Expression (1). In the present embodiment, the initial feature quantity score vector is a vector having the same length as the feature quantity list and all values being zero.

The calculation unit 103 generates an initial feature quantity correlation matrix for storing correlation values between feature quantities (SS3121). Formula (2) shows an example of the initial feature quantity correlation matrix. In this embodiment, the initial feature quantity correlation matrix is a matrix in which the number of rows and the number of columns are both equal to the length of the feature quantity list and all values are zero.

  The calculation unit 103 determines whether there is a feature quantity that has not been selected in the feature quantity list (S3122). If all the feature quantities have not been selected, the process proceeds to S3123, and if selected, the process proceeds to S3128. . If not all feature amounts have been selected, a feature amount that has not yet been selected from the feature amount list is selected as a selected feature amount (S3123).

  The calculation unit 103 determines whether there is an unselected viewing history in the online video viewing history (S3124). If not all viewing history has been selected, the process proceeds to S3125, and if selected, the process proceeds to S3122. move on. A viewing history that has not yet been selected from the online video viewing history is selected as a selected viewing history (S3125).

The calculation unit 103 updates the feature amount score vector with the feature amount of the selected viewing history (S3126). When the selected feature amount is included in the selected viewing history, addition is performed to the component of the feature amount score vector corresponding to the selected feature amount. In this embodiment, when the selected feature amount is included in the selected viewing history, 2 is added when the attribute of the selected feature amount is a search word, and the selected feature amount is selected when included in the selected viewing history. It is assumed that 1 is added when the attribute of the feature quantity is a search word, and no addition is performed when the selected feature quantity is not included in the selected viewing history. It suffices if a difference occurs in the components of the score vector, and the numerical value of the added value may be corrected as appropriate. Formula (3) shows an example of updating the feature value score vector. If “Singer A Sky PV” in the example of Table 3 is selected as the selected viewing history and “Singer A” is selected as the selected feature amount, 2 is added to the value corresponding to singer A by updating the feature amount score vector. The It is also possible to use the viewing state of the selected viewing history as the added score. In this case, 2 is added for “completed”, and 1 is added for “suspended”.

The calculation unit 103 updates the feature amount correlation matrix with the feature amount of the selected viewing history (S3127). When the selected feature amount is included in the selected viewing history, attention is paid to the component of the row corresponding to the selected feature amount, and addition is performed to the component of the column corresponding to all the feature amounts included in the selected viewing history. In the present embodiment, it is assumed that 1 is added to the component of the column corresponding to all the feature amounts included in the selected viewing history. Formula 4 shows an example of updating the feature amount correlation matrix. Assuming that “Singer A Sky PV” in the example of Table 3 is selected as the selected viewing history and “Singer A” is selected as the selected feature quantity, “Singer A” “Sky” among the components of the row corresponding to “Singer A” "1" is added to the component corresponding to "PV".

The calculation unit 103 normalizes the values of the feature amount score vector and the feature amount correlation matrix, assigns the feature amount score vector to the feature amount list, and sorts the scored feature amount list in descending order of score (S3128). Expression (5) is an example of normalization of the feature amount score vector.

Here, score_i is a normalized score of the feature value F_i, and N_i is a value before normalization of the feature value score vector of the feature value F_i. In this embodiment, the feature amount score is normalized to 0-1 according to equation (5). The normalized feature quantity score is assigned to the corresponding feature quantity in the feature quantity list. Expression (6) is an example of normalization of the feature amount correlation matrix.

Where rel_ {i, j} is the (i, j) component of the normalized feature correlation matrix, and R_ {i, j} is the value before normalization of the (i, j) component of the feature correlation matrix is there. Table 5 is an example of a scored feature quantity list obtained in S312. In Table 5, the score obtained for each feature amount is given. Table 6 shows an example of the feature amount correlation matrix obtained by the feature amount evaluation information calculation process of S312. In Table 6, values representing correlations with other feature amounts are stored in rows corresponding to the feature amounts.

  FIG. 8 is a flowchart showing a specific example of the feature value set generation processing in S313.

  The calculation unit 103 determines whether or not there is a feature quantity not selected in the scored feature quantity list (S3130). If all the feature quantities have not been selected, the process proceeds to S3131. The process ends. If not all feature quantities have been selected, a feature quantity that has not yet been selected from the scored feature quantity list is selected as a selected feature quantity (S3131). In this embodiment, “Singer A” in Table 5 is selected as the selected feature amount in the first loop.

  The calculation unit 103 sets an initial correlation flag indicating whether or not there is a correlation (S3132). In this embodiment, the value of the initial correlation flag is set to 0, assuming that the correlation flag is 0 if it is invalid and 1 if it is valid.

  The calculation unit 103 determines whether there is a feature quantity set that has not been selected from among the feature quantity sets (S3133). If all feature quantity sets have not been selected, the process proceeds to S3134. Proceed to Since no feature quantity set has been generated in the first loop, it is determined that all feature quantity sets have been selected, and the process advances to S3139.

  The calculation unit 103 selects a feature value set that has not yet been selected from the feature value set as a selected feature value set (S3134). In this embodiment, it is assumed that the feature amount set 1 is selected as the selected feature amount in the second calculation.

  Next, a correlation value between the selected feature value and the selected feature value set is calculated (S3135). A feature quantity correlation matrix is used for the calculation of the correlation value. For example, a value is obtained by referring to a row corresponding to the selected feature value and a column component corresponding to the first feature value of the selected feature value set in the feature value correlation matrix. In this embodiment, “Sky” is selected as the selected feature value, and “Singer A” of feature value set 1 is selected as the first feature value of the selected feature value set. 0.4 corresponding to the correlation between “Sky” (F2) and “Singer A” (F1) is extracted as a correlation value.

  The calculation unit 103 determines whether or not the correlation value satisfies the criterion (S3136). If the criterion is satisfied, the process proceeds to S3137. If the criterion is not satisfied, the process proceeds to S3133. In this embodiment, it is assumed that the criterion is satisfied if the correlation value is not zero. Since the correlation value of “sky” is 0.4, the criterion is satisfied, and the selected feature value is added to the selected feature value set (S3137). In this embodiment, “sky” is added to the feature amount set 1. When the selected feature amount is added to the feature amount set, the value of the correlation flag is validated (S3138). Here, it is valid when the correlation flag is set to 1, and is invalid when it is not 1.

  When the selected feature amount is added to the feature amount set or when all the feature amounts have been selected, the calculation unit 103 determines whether or not the value of the correlation flag is valid (S3139). The process proceeds to S3130, and if invalid, the process proceeds to S31310. Since the correlation flag is invalid in the first calculation, the process proceeds to S31310.

  The calculation unit 31 generates a new feature amount set using the selected feature amount (S31310). In this embodiment, since “Singer A” is the selected feature amount in the first loop, the feature amount set 1 having “Singer A” as the first feature amount is generated.

  In this embodiment, after returning to S3130, “sky” is selected as the selected feature amount in the second loop. When “sky” is the selected feature value, the feature value set 1 exists as a feature value set not selected in S3133.

  As described above, by repeating the feature value set generation process of S313, the feature value set calculation unit automatically generates a feature value set. FIG. 9 shows an example of the generated feature amount set. Each feature amount set includes a first feature amount having a high score and a feature amount having a correlation with the first feature amount. The calculation unit 103 stores the generated feature amount set in the first storage unit 107 in S314.

  FIG. 10 is a flowchart illustrating a specific example of processing in the determination unit 104. The determination unit 104 determines the association between the feature quantity set and the broadcast program information, and creates a related program list including the program ID of the related broadcast program, the feature quantity set ID, and the upper feature quantity association degree.

  The determination unit 104 creates an empty initial related program list (S330).

  The determination unit 104 determines whether there is broadcast program information that has not been selected in the broadcast program information (S331). If all the broadcast program information has not been selected, the process proceeds to S332. Exit. If it is not yet selected from the broadcast program information, one unselected broadcast program information is selected as selected broadcast program information (S332).

The determination unit 104 determines whether there is a feature amount set that is not selected in the feature amount set (S333). If all feature amount sets have not been selected, the process proceeds to S334, and if selected, S331 is performed. Proceed to If not yet selected from the feature value set, one unselected feature value set is selected as a selected feature value set (S334). The determination unit 104 calculates the upper feature quantity of the selected feature quantity set and the upper feature quantity relevance of the selected feature quantity (S335). In this embodiment, the higher-level feature quantity relevance is expressed by Equation 7.

Here, S is the selected feature value set, T is the selected broadcast program information, rel (S, T) is the degree of relevance of the upper feature values of S and T, f_ {S, i} is the i-th feature value of S, [f_T] Is a set of feature quantities included in T, [f_S] _k is any feature quantity from the first feature quantity to the k-th feature quantity of the selected feature quantity set, and score (f_S, i) is the feature quantity f_S, i It is a score. In this embodiment, a keyword is used as a feature amount of broadcast program information. Equation 7 calculates the upper feature quantity relevance if at least one of the first to k-th feature quantities in the selected feature set is included in the selected broadcast program information, and if not, This indicates that the relevance value is set to 0. In this embodiment, it is assumed that the feature amount set ID1 in FIG. 9 is selected as the selected feature amount set, and the program ID 20100321sfg in Table 1 is selected as the selected broadcast program information. Also assume that k = 1. At this time, [f_S] _k of the selected feature set is [“Singer A”], and the feature set [f_T] of the selected broadcast program information is [“Singer A”, “Singer F”, “New song”, “Sky”. , “Show”, “Live”,…]. Since the element “singer A” of [f_S] _k is included in [f_T], the determination unit 104 calculates the upper feature amount relevance degree according to Equation 7. Here, since “Singer A”, “Sky”, and “New song” are included in the selected broadcast program among the selected feature value sets, the higher-level feature value relevance is 0.95 + 0.75 + 0.54 = 2.24.

  The determination unit 104 determines whether or not the calculated upper feature quantity relevance satisfies the standard (S336). If the standard is satisfied, the process proceeds to S337 because the selected broadcast program is related, and if the standard is not satisfied, the process proceeds to S333. . The reference can be, for example, rel (S, T)> 0. At this time, if the upper feature amount of the selected feature amount set is included in the selected broadcast program information, the criterion is satisfied. In this embodiment, it is determined that the program ID 20100321sfg in Table 1 satisfies the criteria. If the standard is satisfied, the related program list is updated as related (S337). The related program list includes a program ID of a broadcast program that is determined to be related, a related feature value set ID, and a value of a higher feature value relevance level. In this embodiment, 20100321sfg, ID1, and 2.24 are added.

Through the above processing, the determination unit 104 automatically generates a related program list. Table 7 is an example of a related program list.

  When using the higher-level feature value relevance level, only programs that are more valuable to the user can be set as related programs. For example, a method of considering a broadcast program as a related program when one of the feature quantities is simply included in the broadcast program information without using the upper feature quantity of the feature quantity set is considered.

  At this time, for example, the second feature amount “sky” of the feature amount set ID1 in FIG. 8 and the fourth feature amount “nabe” of ID2 are “world news” of the program ID 2003100301wx0 of the broadcast program information in Table 1 and “today” of the program ID 2003100302fsa. Are also included in For the user set here, “Sky” co-occurs with “Singer A”, and “Nabe” co-occurs with “Cat”, thereby increasing the value. Co-occurrence information cannot be used for simple word matching, and in the “World News” of the former program ID 2003100301wx0, “Sky” appears in the context of “Sky Tower”, which is a building. Selected. In the latter case, the program ID 2003100302fsa is selected even though it appears in the context of “nabe” which is food in “Today's dinner”. Thus, it is determined that broadcast programs that are not actually related are also related. In the present embodiment, by using the correlation between feature quantities and using the higher-order feature quantity relevance level, only programs that are valuable to the user can be set as related programs.

  FIG. 11 is a flowchart showing a specific example of processing in the updating unit 34. The update unit 105 generates a search term set, and automatically updates the search history of content that the user is interested in by transmitting each element of the search term set to the search information acquisition unit of the online video viewing control unit 11. .

  First, the update unit 105 generates an empty search term set (S340).

  Next, the update unit 105 determines whether or not there is a related program ID that is not selected in the related program list (S3401). If all the related program IDs are not selected, the process proceeds to S342 and has been selected. If so, the process proceeds to S347. When all the related program IDs are not selected, one related program ID that has not been selected from the related program IDs is selected as the selected related program ID (S342). In this embodiment, it is assumed that the related program ID 20100321sfg is selected in Table 1.

  The update unit 105 reads the broadcast program information corresponding to the selected related program ID from the broadcast program information storage unit 12 (S343). In this embodiment, the information of “music station” of the program ID 20100321sfg in Table 1 is read. The updating unit 105 selects one related feature amount set corresponding to the selected related program ID in the related program list as a selected related set (S344). In this embodiment, the related feature quantity set ID1 corresponding to the selected related program ID is selected from the related program list in Table 7.

  The update unit 105 reads the feature value set corresponding to the selected feature value set ID from the feature value set first storage unit 107 (S345). In this embodiment, the feature amount set ID1 in FIG. 8 is read. A search word is added to the search word set from the selection related program and the selected feature amount set (S346). The search word is composed of a combination of words included in the selection related program in the selection feature amount set. In this embodiment, since “Singer A”, “Sky”, and “New song” are included in the selection related program in the feature amount set ID1, a search word that includes these three words as a set is added.

  The update unit 105 transmits each search term to the search information acquisition unit of the network video viewing control unit 11 for the added search term set (S347). In this embodiment, “Singer A_Sky_New Song” is transmitted. As described above, the search information acquisition unit of the net video viewing control unit 11 stores a new search result in the net video viewing history storage unit 12.

  Through the above processing, the update unit 105 automatically searches for a net video that is of interest to the user and updates the history. When the user desires to view a new video about the singer A who is interested in the viewing of the broadcast program, the user can save time and effort for the user to re-search for the latest information by the update unit 105.

Table 8 shows an example of a search term set generated by the update unit 105. Since only the words common to the feature amount set and the related program are added to the search word set, the search word generated from the related program ID 20100321veq in Table 1 is only “cat”.

  FIG. 12 shows an example of a screen that the display unit 106 outputs to the user. In the example, in an electronic program guide of a normal broadcast program, related programs are presented in different colors (dot pattern in FIG. 12). In FIG. 12, the color of the related program is changed, but it may be changed if the presentation method is easy to understand for the user, such as making the frame of the program table thicker than usual. Furthermore, a related net moving image display unit 350 may be provided that reads information on a moving image including a feature amount set that is a basis for determining a related program from the net moving image viewing history storage unit 12 and displays the information simultaneously with the related program. The feature set that is the basis for determining the related program is highlighted with bold and underlined, but as with the program guide display, the display should be easy to understand for the user, and the color can be changed. But you can. In addition, the search is updated by the update unit 105 for a moving image that the user is interested in among the related programs.

  In this embodiment, the related net video information display unit displays the net video information including the information of the latest search list searched for “Singer A_Sky_New Song” in addition to the video previously watched about “Singer A”. Is done.

  According to the present embodiment, it is possible to provide a content recommendation device capable of presenting content that matches user preferences.

  The content recommendation device according to the present embodiment is a communication I / F that communicates with a control device such as a CPU (Central Processing Unit) and a storage device such as a ROM (Read Only Memory) and a RAM connected to a network or the like. An external storage device such as an HDD (Hard Disk Drive) or a CD (Compact Disc) drive device, a display device such as a display device that presents for content recommendation, and an operator for identification processing and result acquisition It has a hardware configuration that uses an ordinary computer and an input device such as a keyboard and mouse, and a bus that connects each part.

  The recommendation processing program executed by the content recommendation device according to the present embodiment is a file in an installable or executable format, and is a CD-ROM (Compact Disk Read Only Memory), a flexible disk (FD), a CD-R ( The present invention can be provided by being recorded on a computer-readable recording medium such as a Compact Disk Recordable (DVD) or a DVD (Digital Versatile Disk).

  Further, the recommendation processing program executed by the content recommendation device according to the present embodiment may be stored on a computer connected to a network such as the Internet and provided by being downloaded via the network. . The processing program executed by the content recommendation processing apparatus according to the present embodiment may be provided or distributed via a network such as the Internet.

  Further, the processing program of the present embodiment may be provided by being incorporated in advance in a ROM or the like.

It should be noted that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

DESCRIPTION OF SYMBOLS 101 ... 1st acquisition part, 102 ... 2nd acquisition part, 103 ... Calculation part, 104 ... Determination part, 105 ... Update part, 106 ... Display part, 1st Storage unit ... 107, second storage unit ... 108

Claims (7)

A first acquisition unit that acquires first meta information included in the first content viewed by the user and the viewing history of the user;
A second acquisition unit that acquires second meta information included in the second content to which the structured information is added;
A calculation unit that calculates a feature value set using a score of the feature value obtained from the viewing history and a correlation between the feature values;
A determination unit that determines whether or not the first content and the second content included in the viewing history are related using the feature amount set;
A content recommendation device comprising: a display unit that displays the second content using a determination result of the determination unit. A first storage unit for storing the feature quantity set;
The calculation unit selects the feature amount in descending order of score, calculates a correlation with a second feature amount in the feature amount set stored in the storage unit,
When the correlation with the second feature quantity is high, the feature quantity is added to the feature quantity set, and when the correlation with the second feature quantity is low, a new feature quantity set including the feature quantity is added. The content recommendation device according to claim 1, wherein the content recommendation device is generated. A second storage unit for storing the first content viewing history;
When the second content is related to the first content, it further includes an updating unit that searches the first content using the feature amount set and updates the viewing history. The content recommendation device according to claim 1 or 2.   The display unit displays the first content and the second content together when the second content is related to the first content. The content recommendation device according to any one of the preceding claims.   5. The content recommendation device according to claim 1, wherein the determination unit determines using a feature amount having a score equal to or higher than a predetermined threshold included in the feature amount set.   6. The content recommendation apparatus according to claim 3, wherein the update unit searches for content in a database connected to a network and acquires new content.   The content recommendation apparatus according to claim 6, wherein the display unit displays the new content read by the updating unit.

Images