JP2014119993A - Content recommendation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a content recommendation device capable of recommending contents matching user's needs when contents are searched for by grouping users on the basis of similarities of time-series transition of content evaluation and receiving a recommendation for a content to be viewed next from a similar user group.SOLUTION: A content recommendation device includes content viewing data reception means of receiving content viewing data from each content viewing device, viewer group generation means of generating a viewer group on the basis of the received content viewing data, and content recommendation means of determining a recommendation content on the basis of a content viewing history or current viewing state of a viewer in the generated viewer group, and reporting the recommendation content to the members in the group. The viewer grouping means puts viewers who have the same set of successive viewing contents in the same group.

Description

  The present invention relates to a content recommendation device.

  In recent years, contents are not limited to those distributed from broadcasting stations, and for example, a large amount of contents are distributed including a content distribution service in a network. The enrichment of content enables users to view content that is closer to their tastes, but on the other hand, there is too much content and a new problem has arisen that appropriate content cannot be selected.

  Here, as a method for selecting content close to the user's preference from many contents, a method of using evaluation values of other users is known.

  For example, as shown in Patent Literature 1, an evaluation value given by each user for content is acquired, the number of contents evaluated by the user, an evaluation value given to each content by the user, and content evaluated by the user Using at least one of the evaluation number given by each user and the evaluation value given by each user for the content evaluated by the user, a user statistic indicating a user's evaluation tendency is calculated and the user is calculated. An invention for controlling the presentation of information related to content to a user based on a statistic is known.

JP 2009-157907 A

  However, when content is recommended to a user simply based on content evaluation, there is a problem that appropriate content is not necessarily recommended. For example, in the above-mentioned patent document 1, since recommendation is made based on the evaluation history in units of content, it depends on the content length / density, the change in mood while the user is viewing, etc. It is not possible to obtain temporal changes in evaluation history. That is, since content recommendation based on information on how the user views the content cannot be performed, there has been a problem that appropriate content is not recommended.

  Furthermore, in the user group based on the accumulation of the evaluation history of Patent Document 1 described above, there has been a problem that it is not possible to make a recommendation corresponding to the fluid needs of the user when searching for content.

  In view of the above-described problems, the present invention aims to perform user grouping based on the similarity of time-series transition of content evaluation, and receive recommendation of content to be viewed next from a similar user group, To provide a content recommendation device capable of recommending content that matches a user's needs at the time of content search.

In view of the above-described problems, the content recommendation device of the present invention is
Content viewing data receiving means for receiving content viewing data from each content viewing device;
Viewer group generation means for generating a viewer group based on the received content viewing data;
Content recommendation means for determining recommended content based on the content viewing history or current viewing state of the viewer of the viewer group generated and notifying members in the group;
With
The viewer grouping means generates viewers having the same set of continuous viewing content as the same group.

  According to the present invention, a viewer group is generated based on the content viewing data received from each content viewing device, and the recommended content based on the content viewing history or the current viewing state of the viewer of the generated viewer group. And notify members in the group. At this time, it is characterized in that viewers with matching sets of continuous viewing content are generated as the same group. Thereby, viewers with similar tendencies and preferences can be grouped, and content can be recommended based on the group.

  Accordingly, since the content is based on the viewing history of viewers who have similar trends and preferences, the content that the viewer quotients is recommended, and the content can be selected efficiently and appropriately.

It is a figure for demonstrating the whole structure of a content recommendation system. It is a figure for demonstrating the function structure of the content display apparatus in this embodiment. It is a figure which shows an example of the data structure memorize | stored in the memory | storage part of the content display apparatus in this embodiment. It is a figure which shows an example of the data structure memorize | stored in the memory | storage part of the content display apparatus in this embodiment. It is a figure for demonstrating the function structure of the content recommendation server in this embodiment. It is a figure which shows an example of the data structure of the user management table in this embodiment. It is a figure which shows an example of the data structure of the content management table in this embodiment. It is a figure for demonstrating the flow of the whole system in this embodiment. It is an operation | movement flow for demonstrating operation | movement of the content display apparatus in this embodiment. It is an operation | movement flow for demonstrating operation | movement of the content recommendation server in this embodiment. It is an operation | movement flow for demonstrating the user operation information total process in this embodiment. It is an operation | movement flow for demonstrating the user group creation process in this embodiment. It is an operation | movement flow for demonstrating the information total process in a group in this embodiment. It is an operation | movement flow for demonstrating the recommendation content extraction process in this embodiment. It is a figure for demonstrating the operation | movement in this embodiment. It is a figure for demonstrating the operation | movement in this embodiment. It is a figure for demonstrating the operation | movement in this embodiment. It is a figure for demonstrating the operation | movement in this embodiment.

  Next, a content recommendation system when the present invention is applied will be described. In this embodiment, a digital television included in the content recommendation system will be described as an example. The present embodiment is an example for understanding the invention, and the present invention is not limited to the embodiment and is not applied.

[1. overall structure]
First, the overall configuration of the system will be described. FIG. 1 is a diagram showing the entire content recommendation system 1000. In the content recommendation system 1000, the content display device 1 is connected to a network. In addition, the content recommendation server 3 is connected to the system 1000.

  The content display device 1 is a device that can reproduce and display content distributed by the content recommendation server 3. The content recommendation server 3 is a server that can recommend content to be distributed. Here, the content in the present embodiment will be described as being distributed from the content recommendation server 3 for convenience of explanation. However, for example, other content distribution devices may be connected, or other content distribution providers It is good also as referring to the content distributed.

[2. Functional configuration]
Next, functional configurations of the content display device 1 and the content recommendation server 3 will be described with reference to the drawings.

[2.1 Content display device]
First, the functional configuration of the content display device 1 will be described with reference to FIG. As shown in FIG. 2, the content display device 1 includes a control unit 100, a content reception unit 110, a content presentation control unit 120, an information input unit 130, an input information analysis unit 140, a storage unit 150, a communication A unit 160, an operation unit 170, a display control unit 180, and an audio output unit 185 are configured.

  The control unit 100 is a functional unit for controlling the entire content display device 1. The control unit 100 realizes various functions by reading and executing various programs stored in the storage unit 150, and is configured by, for example, a CPU (Central Process Unit).

  The content presentation control unit 120 is a control unit for displaying content presented from the outside. For example, the content recommended by the content recommendation server 3 is presented / displayed and selected by the user. The selected content is displayed on the content display device 1.

  The information input unit 130 is a functional unit to which an evaluation from a user is input. For example, in the present embodiment, content evaluation is performed by inputting an evaluation when the user likes the content display (for example, touching a “Like” button displayed on the display screen). Can be done. The input evaluation is analyzed by the input information analysis unit 140 and transmitted to the content recommendation server 3 as evaluation data.

  In the case of this embodiment, the input number of evaluations is totaled every predetermined time, and the totaled data is used as content evaluation data. In addition, you may process by using the information at the time of evaluation as evaluation data.

  The storage unit 150 is a functional unit that stores various programs and various data necessary for the operation of the content display device 1. The storage unit 150 includes, for example, a semiconductor memory, an HDD (Hard Disk Drive), or the like.

  In the present embodiment, content viewing start time data 152, evaluation data 154, and content viewing end time data 156 are stored.

  Here, an example of the content viewing start time data 152 is shown in FIG. 3A, an example of the evaluation data 154 is shown in FIG. 3B, and an example of the content viewing end time data 156 is shown in FIG.

  That is, the content viewing start time includes the user ID (for example, “U1”), the content ID (for example, “C1”), and the viewing start time as content viewing start time data 152 as shown in FIG. The time (for example, “2012/06/05 19:33:00”) is stored in association with each other.

  Further, as shown in FIG. 3C, the content viewing end time is the user ID (for example, “U1”), the content ID (for example, “C1”), and the viewing end time. The time (for example, “2012/06/05 19:56:00”) is stored.

  Data evaluated by the user is stored as evaluation data 154. For example, as shown in FIG. 3B, a user ID (eg, “U1”), a content ID (eg, “C1”), a predetermined time interval ID (eg, “T1”), and the time The section evaluation number (for example, “1”) included in is stored.

  Note that the evaluation data 154 may be aggregated for each section as shown in FIG. 3B, or is initially associated with raw data (for example, a user ID and a content ID as shown in FIG. 4). And the evaluated time) may be stored. A predetermined section evaluation number can be calculated from the raw data of the evaluated time.

  The communication unit 160 is a functional unit for performing data communication with other devices. For example, it is configured by an interface that can be connected by a wired connection such as Ethernet (registered trademark), a wireless LAN such as IEEE802.11a / b / g / n, a wireless connection method such as a WAN, 3G line, or WiMAX.

  The operation unit 170 is a functional unit for the user to perform various operations on the content display device 1. For example, it is configured by a touch panel, hardware keys, a remote controller, and the like.

  The display control unit 180 and the audio output unit 185 are functional units for outputting video and audio of content. The display control unit 180 may be, for example, a display device such as an LCD connected to the outside, or the display control unit 180 may function as a display unit. Also, the audio output unit 185 may output audio to a speaker or the like connected to the outside, or the audio output unit 185 itself may output audio.

[2.2 Content recommendation server]
Next, the functional configuration of the content recommendation server 3 will be described with reference to FIG. As shown in FIG. 5, the content recommendation server 3 includes a control unit 200, a communication unit 210, a user operation information totaling unit 220, an in-group information statistics unit 230, a recommended content extraction unit 240, and a recommended content transmission unit. 250 and a storage unit 260.

  The control unit 200 is a functional unit for controlling the entire content recommendation server 3. The control unit 200 implements various functions by reading and executing various programs stored in the storage unit 260, and is configured by, for example, a CPU (Central Process Unit).

  The communication unit 210 is a functional unit for communicating data and the like with other devices. For example, it is configured by an interface that can be connected by a wired connection such as Ethernet (registered trademark), a wireless LAN such as IEEE802.11a / b / g / n, a wireless connection method such as a WAN, 3G line, or WiMAX.

  The user operation information totaling unit 220 is a functional unit for totaling user operation information (for example, evaluation data) transmitted from each content display device connected to the content recommendation server 3. Specifically, this is realized by executing user operation information aggregation processing described later.

  The in-group information statistics unit 230 is a functional unit that aggregates information in a user group and creates a group. Specifically, this is realized by executing an intra-group information aggregation process described later.

  The recommended content extraction unit 240 is a functional unit for extracting content suitable for the user from the content. Specifically, it is realized by executing recommended content extraction processing described later. Then, the recommended content (or information regarding the recommended content) extracted by the recommended content extraction unit 240 is transmitted by the recommended content transmission unit 250.

  The storage unit 260 is a functional unit that stores various programs and various data necessary for the operation of the content recommendation server 3. The storage unit 260 includes, for example, a semiconductor memory, an HDD (Hard Disk Drive), or the like.

  In addition, the storage unit 260 stores a user management table 262 and a content management table 264.

  The user management table 262 is a table for managing users. An example of the data structure of the user management table 262 is shown in FIG. As illustrated in FIG. 6, the user management table 262 stores a content ID and a viewing end time of the content ID in association with the user ID. Furthermore, evaluation data for each content is stored as the number of evaluation data for each time interval ID.

The content management table 264 is a table for managing content. The content management table 264 is a table in which various information is stored in association with the content ID, and an example thereof is shown in FIG. In this embodiment,
(1) A table in which the user ID that viewed the content corresponding to the content ID and the viewing end time are stored (FIG. 7A).
(2) The ID of the content (preceding viewing content) reproduced immediately before the content with the content ID is viewed, the viewing end time of the preceding viewing content, and the viewing user ID viewed at the viewing end time. Stored table (FIG. 7B)
(3) The ID of the content (subsequent viewing content) reproduced immediately after the content with the content ID is viewed, the viewing end time of the subsequent viewing content, and the viewing user ID viewed at the viewing end time. Stored table (FIG. 7C)
It is included.

[3. Process flow]
[3.1 Overall system flow]
First, the flow of the entire system will be described with reference to FIG. In the example of FIG. 8, a case where the contents C1, C2, and C3 are selected and reproduced / displayed will be described. In addition, FIG. 8 demonstrates the flow of the whole process, and the detailed flow of each process is mentioned later.

  First, viewing of the content C1 is started on the content display device 1 (S10). Here, when viewing of the content C1 is started, content viewing start time data is transmitted from the content display device 1 as a client to the content recommendation server 3 (S60). Upon receiving the content viewing start time data, the content recommendation server 3 updates the user management table and the content management table (S30).

  Subsequently, the content display device 1 creates evaluation data of the content C1 while displaying the content (S12). When the viewing of the content C1 ends (S14), the evaluation data and the content viewing end time data are transmitted to the content recommendation server 3 (S62). Thereby, the content recommendation server 3 updates the user management table and the content management table (S32).

  Subsequently, the user starts viewing the content C2 on the content display device 1 (S16). When viewing of the content C2 is started, the viewing start time of the content C2 is transmitted to the content recommendation server as content viewing start time data (S64). Thereby, the content recommendation server 3 updates the user management table and the content management table (S34).

  Subsequently, the content display device 1 creates evaluation data of the content C2 while displaying the content (S18). When the viewing of the content C2 ends (S20), the evaluation data and the content viewing end time data are transmitted to the content recommendation server 3 (S66). Thereby, the content recommendation server 3 updates the user management table and the content management table (S36).

  Subsequently, the content recommendation server 3 executes a user group creation process (S38) and an in-group information aggregation process (S40). Then, by executing the recommended content extraction process (S42), the recommended content is extracted, and the recommended content data is transmitted to the content display device 1 (S68).

  If there is a recommended content change request in the content display device 1 (S22; Yes), the recommended content change request is transmitted to the content recommendation server 3 (S70). Upon receiving the recommended content change request, the content recommendation server 3 executes recommended content extraction processing again (S44), and transmits newly extracted recommended content data to the content display device 1 (S72).

  The content display device 1 can start viewing the recommended content based on the recommended content data (S24). Note that, similarly to the above-described case, when content viewing is started, content viewing start data is transmitted (S74).

[3.2 Processing of content display device]
Next, the operation of the content display device 1 will be described with reference to FIG. First, when content to be viewed (reproduced) is selected (step S102), content viewing start time data is transmitted to the content recommendation server 3 (step S104).

  Subsequently, when user operation information is acquired (step S106), evaluation data is created (step S108). Here, the user operation information in the present embodiment executes, for example, a process of counting the number of times when a user inputs an evaluation while viewing content, every predetermined time. In this embodiment, a positive evaluation is input as an evaluation (that is, an operation is performed when the viewer thinks that the viewer thinks “good”), but a negative evaluation is input ( That is, the operation may be performed when the viewer thinks that it is “bad”), or both may be input.

  Here, when the viewing of the content is terminated (step S110; Yes), the evaluation data for the content generated in step S108 and the content viewing end time data are transmitted to the content recommendation server 3 (step S112).

  Here, when the information regarding the recommended content is received from the content recommendation server 3 (step S114; Yes), the recommended content is presented (step S116). Here, when the request for changing the presented recommended content is made by the user (step S118; Yes), the recommended content request is transmitted to the content recommendation server 3 (step S122).

  Further, when the request for changing the recommended content is not made (step S118; No), the processing for selecting the content and viewing it is continuously executed (step S120).

[3.3 Processing of content recommendation server]
Next, the operation of the content recommendation server 3 will be described with reference to FIG. First, it is determined whether information is received from the content display device 1 (terminal) (step S202). Specifically, content viewing start time data, content end time data, evaluation data, and the like are received from the content display device 1.

  Subsequently, a user operation information totaling process is executed (step S204). Here, the user operation information aggregation process will be described in detail with reference to FIG. The time interval between the time when the current content of the same user is selected (content viewing start time) and the time when viewing of the immediately preceding content is terminated (the current content viewing start time and the content viewing end time of the immediately preceding content data) It is determined whether or not the time is equal to or less than a predetermined threshold (Δt or less) (step S206).

  Note that Δt is an upper limit value of a time interval in which the user can consider that the user has continuously viewed two contents, and may be within a range of “immediately” or “immediately on the spot”. It is conceivable to set a fixed value such as 1 minute or within 10 minutes.

  In addition, it is possible to set a time unit that can be regarded as a continuation of the same mood in consideration of a user's life pattern and biorhythm. For example, continuous content such as continuous dramas and daily news programs can also be detected as continuous programs viewed in the same mood based on time intervals.

  In this case, by setting Δt so as to match the broadcast interval of continuous content, it is possible to determine whether or not the content is continuous viewing content depending on whether or not it matches. As a result, a user who starts watching in the usual mood at the usual time can receive content recommendation considering the connection from the previous viewing.

  If Δt is equal to or smaller than the predetermined threshold (step S206; Yes), a user group is created (step S208), and the in-group information aggregation process is executed based on the created user group (step S210). ).

  By responding to usage scenes in which viewing at a specific location is customarily performed, the viewpoint of location can be used to determine whether to continuously view content. For example, a user who regularly watches on a train views the previous time on the train regardless of the viewing time interval by turning on the setting that viewing at a predetermined viewing location (train) is regarded as continuous viewing. It can be determined as continuous viewing of content. As a result, even for a user who regularly watches in the usual place and in the usual mood, the content recommendation considering the connection with the previous viewing can be received.

  Regarding determination of continuous content viewing, implementation other than automatic determination by the system based on set conditions may be considered. Specifically, after the user has selected the content to be viewed, before playback, the user may explicitly present a question sentence such as “Are you viewing the content that was previously viewed?” It is conceivable that the “new viewing” can be designated and input manually by the user such as a pop-up menu or a button, and whether or not to continue viewing.

  As described above, when it is determined that the user has watched two contents in succession, the content recommendation server 3 executes recommended content extraction processing (step S212), and transmits information related to the recommended content to the content display device 1. (Step S214). Here, when the recommended content request is received, the recommended content extraction process is executed again (step S216; Yes → step S212).

[3.3.1 User operation information aggregation process]
User operation information aggregation processing will be described with reference to FIG. First, when content viewing start time data is received (step S1002), a user ID included in the content viewing start time data is extracted (step S1004).

  At this time, when the time (content interval time) since the content was viewed immediately before in the user ID is less than the predetermined time interval (Δt), the content viewed immediately before the user with the user ID is displayed. The content management table 264 that manages the content that has now been viewed is stored as the preceding viewing content (step S1008).

  Subsequently, evaluation data for each time zone on the content is received (step S1010). Finally, when content viewing end time data is received (step S1012; Yes), the user management table 262 and the currently viewed content are displayed. The management table 264 is updated (step S1014).

  Specifically, the user management table 262 stores the current content viewing end time and evaluation data (the number of evaluation data for each time interval) in association with the user ID and the content ID.

  Further, the current content management table 264 stores the viewing end time in association with the currently viewed content ID and the user ID, and information related to the preceding viewing content that is the content viewed immediately before (preceding viewing content ID, viewing ID). End time and viewing user ID) are stored.

  Further, the subsequent content ID, the viewing end time, and the viewing user ID are stored in the preceding content management table with the currently viewed content as the subsequent viewing content.

[3.3.2 User group creation process]
Next, user group creation processing will be described with reference to FIG. First, it is determined whether or not the preceding viewing content corresponding to the current content is stored in the content management table 264 (step S2002).

  When the preceding viewing content is stored (step S2002; Yes), the ID of the user who has selected the channel for the current viewing content is extracted from the content management table 264 as a continuation of the preceding viewing content (step S2002). S2004).

  Here, evaluation data corresponding to the user ID extracted in step S2004 is acquired (step S2006). Then, the degree of similarity between users is calculated from the acquired evaluation data, a grouping process is executed (step S2008), and an in-group user ID is extracted (step S2010). Here, an operation example of the grouping process will be described later.

[3.3.3 Group information aggregation process]
Next, intra-group information aggregation processing will be described with reference to FIG. As shown in FIG. 13, first, a user ID in the group is acquired (step S2502). Then, the subsequent viewing content ID stored in the content management table 264 is extracted by filtering with the in-group user ID (step S2504).

[3.3.4 Recommended content extraction processing]
Next, the recommended content extraction process will be described with reference to FIG. First, content with high viewing frequency is extracted (step S3002). The top ranking content is recommended from the extracted content (step S3004).

  Here, when the recommended content change request is received (step S3006; Yes), the content having a characteristic variation is extracted from the viewing frequency (step S3008). Then, the top ranking content is extracted as the recommended content (step S3010).

  If the recommended content change request is received (step S3012; Yes), the ranking of the top instantaneous viewing frequency is extracted (step S3014). Then, the top ranking content is extracted as the recommended content (step S3016).

[4. Example of operation]
In the operation flow described above, specific operations will be described using some specific examples.

[4.1 Create user group]
First, a specific example of creating a user group will be described. Here, a user group is created by the following method.

[4.1.1 First Method]
First, a method for creating a user group from the similarity of evaluation value vectors will be described. FIG. 15 is a diagram showing evaluation data for each user ID. The number of evaluations is recorded for each time segment. FIG. 15 is a diagram showing evaluation data when the first content is C1, the second content is C2, and the transition of the viewing content is C1 → C2.

  FIG. 15A shows evaluation data values. Here, in each section, a change vector of evaluation values is created with “1” being larger than the evaluation value of the previous section and “−1” being smaller. FIG. 15B shows the change vector in this case. Since the end points cannot be subtracted, the dimensions of the evaluation data values of the contents C1 and C2 are reduced one by one, and FIG. 15B becomes an 8-dimensional vector.

  For each vector, the covariance of the two vectors is divided by the product of the standard deviation of each vector to determine the correlation coefficient. Then, users whose correlation coefficient is 0 or more are grouped. In this figure, the similarity between the users U1 and U2 has a correlation coefficient of approximately −0.17408, and is determined not to be in the same group. Further, the similarity between the users U1 and U3 has a correlation coefficient of about 0.700649 and is determined to be the same group.

  As described above, by calculating the increase / decrease of the evaluation value of the adjacent time interval instead of the absolute value of the evaluation data, it is possible to capture the characteristics of the mood change while the user is viewing the content, Can be extracted.

  The above-described method described in the present embodiment is an example of implementing the feature of the present invention in which the similarity between users is determined by capturing the feature of transition (increase / decrease) in the evaluation value of adjacent time intervals. The invention is not limited to the details of this implementation.

[4.1.2 Second method]
Next, a method of calculating the top three evaluation values and determining based on the similarity of the distribution of the sections with the evaluation number “0” will be described.

  First, FIG. 16A shows evaluation data values. Here, among the sections, the sections with the highest three evaluation values and the sections with no evaluation are extracted. Then, the distribution vector is created by setting the upper section and the section where the evaluation data is 0 as “1” and the other sections as “0”. In addition, when evaluation is a tie, it calculates using a younger section number.

  FIG. 16B is a diagram specifically showing distribution vectors. The covariance of two vectors is divided by the product of the standard deviation of each vector to obtain a correlation coefficient. Then, users whose correlation coefficient is 0 or more are grouped.

  For example, the similarity between U1 and U2 has a correlation coefficient of 0.6, and is determined to be the same group. Further, the similarity between U1 and U3 has a correlation coefficient of approximately 0.408248 and is determined to be the same group.

  The above method captures the similarity of the part that the user who watched the content is most impressed or the part that the user has not felt so much by calculating the similarity of the distribution of the characteristic time interval of the evaluation value. Users with similar values can be grouped into the same group.

  Note that the above-described method described in the present embodiment is an example of implementing the feature of the present invention in which the similarity of the distribution of characteristic time intervals of evaluation values is captured and the similarity of preference between users is determined. Yes, the invention is not limited to the details of this implementation.

[4.1.3 Third Method]
Subsequently, the third method will be described. The third method is a method of obtaining a correlation coefficient by dividing the covariance of two vectors by the product of the standard deviation of each vector. Then, a grouping process is performed for users having a correlation coefficient of 0 or more.

  In the diagram shown in FIG. 17, for example, the similarity between U1 and U2 has a correlation coefficient of approximately “0.03066”, and is determined to be the same group. The similarity between U1 and U3 has a correlation coefficient of approximately “0.092375” and is determined to be the same group. Further, the similarity between U1 and U4 has a correlation coefficient of approximately “−0.80001”, and is not determined to be the same group.

  As described above, by adding the time series (evaluation transition for each time interval on the content) to the evaluation of the user's content, the user's preference and mood change at the time of viewing can be captured more accurately. In addition, by updating the grouping of users who have made similar evaluations in a timely manner (at the timing when the two most recent contents are continuously viewed), it is possible to follow users from other users who have viewed with similar preferences and moods. You can receive recommendations for content to watch.

[4.2 Extraction of recommended content]
Next, a specific processing example for extracting recommended content will be described.

[4.2.1 First Method]
First, the first method is a method using history information. That is, from the subsequent viewing content in the group extracted in the group information aggregation process, (1) content with high viewing frequency overall, (2) recent characteristic fluctuations (acceleration increase or high value of viewing frequency) Extract content that has transitions). Here, the viewing end time of the content is used to determine the viewing date and time.

  In addition, the content viewing frequency can be obtained by the number of users who viewed the content in the group / the total number of users in the group in a certain time interval. The time interval is divided by, for example, calculating the viewing frequency at that time in units of one day or one week. The overall viewing frequency is calculated based on all data from the time when counting starts until the present.

For example, in the case of FIG. 18 (a), the current time has passed the time point 5, for example, one day before the time point 5 and two days before the time point 4 is.
(1) Content with the highest overall viewing frequency (R10)
(2) Content with the most recent increase in viewing frequency (R12)
The rate of increase more than 3 times the latest (from time 4 to time 5).
(3) Content whose latest viewing frequency is maintained at a high level (R14)
Viewing frequency of 30% or more in the latest period (time points 3, 4, 5).
Recommended content.

  As described above, by grouping users who have made similar evaluations on a timely basis, and recommending the subsequent content most seen in that group, we recommend content that matches the user's current viewing mood. can do.

  Furthermore, the extraction method 2 that captures recent characteristic fluctuations in the viewing tendency in the user group can perform timely content recommendation reflecting the trend that changes every moment.

  Here, the information regarding the recommended content may be the user himself / herself as the transmission destination, or the user who has not yet viewed the content in the group.

  In addition, as recommendation to an unviewed user, for example, recommendation information from another user may be accumulated and recommendation may be performed at a predetermined timing. For example, recommendations from other users may be accumulated while the power is turned off and notified immediately after the power is turned on, or may be performed as the fourth step of the recommended content extraction process of FIG. The updates from the previous recommendation may be rearranged in order of similarity with the recommendation source user to recommend the higher rank.

  By viewing content recommended by unviewed users in the group, the probability that the evaluation data of the users overlap with each other increases, and there is an effect of improving the accuracy of the recommendation.

[4.2.2 Second method]
As a second method, the current viewing information is used. That is, the most viewed content is recommended within the same user group. For example, a description will be given with reference to FIG. First, the content IDs currently being viewed are totaled. That is, it is a content that has received viewing start time data but has not received viewing end time data.

  Subsequently, the content with the highest instantaneous viewing frequency is extracted in the group. Here, the instantaneous viewing frequency of content can be obtained by the number of users currently viewing in the group at a certain time / the total number of users in the group.

  In the case of FIG. 18B, the highest content C4 of “18%” is recommended. Here, the information regarding the recommended content may be the user himself / herself as the transmission destination, or the user who has not yet viewed the content in the group.

  In addition, as recommendation to an unviewed user, for example, recommendation information from another user may be accumulated and recommendation may be performed at a predetermined timing. For example, recommendations from other users may be accumulated while the power is turned off and notified immediately after the power is turned on, or may be performed as the fourth step of the recommended content extraction process of FIG. The updates from the previous recommendation may be rearranged in order of similarity with the recommendation source user to recommend the higher rank.

[5. Modified example]
The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and the design and the like within the scope of the present invention are also within the scope of the claims. include.

  As a method for determining a set of continuous viewing content, when the viewing end time of the first viewing content and the viewing start time of the second content of the same viewer are within a predetermined time, The case where the viewing location of one viewing content and the viewing location of the second content are within a predetermined range, the case where the viewing content is the first content and the second content designated as continuous viewing content by the viewer, and the like have been described. However, these methods can be realized in combination. That is, it is possible to use a combination of time and place, or to determine whether or not the viewer designates.

  In addition, the program that operates in each device in the embodiment is a program (a program that causes a computer to function) that controls the CPU and the like so as to realize the functions of the above-described embodiments. Information handled by these devices is temporarily stored in a temporary storage device (for example, RAM) at the time of processing, then stored in various ROM or HDD storage devices, and read and corrected by the CPU as necessary. • Writing is performed.

  Here, as a recording medium for storing the program, a semiconductor medium (for example, ROM, a non-volatile memory card, etc.), an optical recording medium / a magneto-optical recording medium (for example, a DVD (Digital Versatile Disc), MO (Magneto Optical), etc. Disc), MD (Mini Disc), CD (Compact Disc), BD, etc.), magnetic recording medium (for example, magnetic tape, flexible disk, etc.), etc. In addition, by executing the loaded program, not only the functions of the above-described embodiment are realized, but also based on the instructions of the program, the processing is performed in cooperation with the operating system or other application programs. The functions of the invention may be realized.

  In addition, when distributing to the market, the program can be stored and distributed in a portable recording medium, or transferred to a server computer connected via a network such as the Internet. In this case, of course, the storage device of the server computer is also included in the present invention.

  In addition, a part or all of each device in the above-described embodiments may be realized as an LSI (Large Scale Integration) that is typically an integrated circuit. Each functional block of each device may be individually chipped, or a part or all of them may be integrated into a chip. Further, the method of circuit integration is not limited to LSI, and may be realized by a dedicated circuit or a general-purpose processor. In addition, when integrated circuit technology that replaces LSI appears due to progress in semiconductor technology, it is of course possible to use an integrated circuit based on this technology.

1: Content display device 3: Content recommendation server 100: Control unit 110: Content reception unit 120: Content presentation control unit 130: Information input unit 140: Input information analysis unit 150: Storage unit 152: Content viewing start time data 154: Evaluation Data 156: Content viewing end time data 160: Communication unit 170: Operation unit 180: Display control unit 185: Audio output unit 200: Control unit 210: Communication unit 220: User operation information totaling unit 230: In-group information statistics unit 240: Recommended content extraction unit 250: Recommended content transmission unit 260: Storage unit 262: User management table 264: Content management table 1000: Content recommendation system

Claims (5)

Content viewing data receiving means for receiving content viewing data from each content viewing device;
Viewer group generation means for generating a viewer group based on the received content viewing data;
Content recommendation means for determining recommended content based on the content viewing history or current viewing state of the viewer of the viewer group generated and notifying members in the group;
With
The content recommendation device, wherein the viewer grouping means generates viewers having the same set of continuous viewing content as the same group. The set of continuous viewing content is
The first content and the second content when the viewing end time of the first viewing content of the same viewer and the viewing start time of the second content are within a predetermined time, the first viewing of the same viewer The first content and the second content or the first content designated as the continuous viewing content by the viewer when the viewing location of the content and the viewing location of the second content are within a predetermined range, and the second content The content recommendation apparatus according to claim 1, wherein the content recommendation device is any one of contents.   The viewer grouping means is a viewer whose evaluation value increases or decreases between adjacent time intervals in the same viewing content, or a viewing in which distribution of time intervals having a characteristic evaluation value is similar in the same viewing content. The content recommendation device according to claim 1 or 2, wherein the users are grouped into the same group.   4. The content recommendation unit determines the recommended content based on the number of views for each viewing section of viewers in the same group or a variation value of the number of views. 5. Content recommendation device.   The content recommendation device according to any one of claims 1 to 3, wherein the content recommendation unit determines a content that is currently viewed most in the same group as a recommended content.

Images