JP5647575B2 - RECOMMENDATION DEVICE, RECOMMENDATION SYSTEM, RECOMMENDATION METHOD, AND PROGRAM - Google Patents

Description

  The present invention outputs a highly accurate recommendation result to a user who is active in providing information, and also outputs a recommendation result from user history information recorded in the recommendation device to a user who is reluctant to provide information. The present invention relates to a recommendation device, a recommendation system, a recommendation method, and a program.

  In recent years, a “recommendation service” has been actively performed in which a user's preference is estimated based on the user's history information and the like, and content that is presumed to match the user's preference is automatically recommended. A typical technique used in such a recommendation service is called “collaborative filtering”. This collaborative filtering defines the similarity between users by comparing user history information such as purchase history and browsing history between each user or between each content, and the user is interested based on the similarity. This is a technique for outputting content predicted to have as a recommendation result. For this collaborative filtering, improved techniques such as automation and high speed are known (for example, see Patent Document 1 and Non-Patent Document 1).

JP 2005-56120 A

P.Resnick, N.Iacovou, M.Suchak, P.Bergstrom, and J.Riedl. “GroupLens: Open Architecture for Copure Copure Cop

  Here, in the technique described in Non-Patent Document 1, in order to analyze history information and generate a recommendation result, it is necessary to accumulate user history information in a centrally managed database. However, there is a problem that some users have a refusal to provide data to the server from the viewpoint of privacy, and the recommendation cannot be provided to users who do not provide such history information.

  With the technology described in Patent Document 1, it is possible to anonymously exchange information by issuing a unique ID based on position and time so that the user can be specified while making it difficult to specify the user. However, it is possible to provide content based on the user's preference. However, even in the above method, the user must provide information, and it is not an effective method for a user who shows a sense of refusal in the act of providing information. In addition, since many recommendation services require login for user identification, there is also a problem that the service cannot be used due to the troublesome task of user registration.

  Therefore, the present invention has been made in view of the above-described problems, and the present invention outputs a highly accurate recommendation result to a user who is active in providing information, and is also responsive to a user who is reluctant to provide information. However, it is an object of the present invention to provide a recommendation device, a recommendation system, a recommendation method, and a program that output a recommendation result from user history information recorded in the recommendation device.

  The present invention proposes the following matters in order to solve the above problems. In addition, in order to make an understanding easy, although the code | symbol corresponding to embodiment of this invention is attached | subjected and demonstrated, it is not limited to this.

  (1) The present invention defines similarity between users based on user history information such as purchase history and browsing history, and recommends content that is predicted to be of interest to the user based on the defined similarity. A recommendation device that uses the collaborative filtering method to output as a result, wherein the purchase history, the browsing history, etc., are published history information storage means for storing the history information of users who are permitted to publish (for example, the publication history information database of FIG. 1) 110), a secret history information storage unit (for example, equivalent to the secret history information database 290 in FIG. 1) for storing history information of users who are not permitted to be disclosed, and similarity calculation from the stored history information. A similarity storage means for storing the calculated similarity (for example, corresponding to the similarity database 130 of FIG. 1), and the public history information storage hand Alternatively, history information acquisition means (for example, equivalent to the history information acquisition unit 210 in FIG. 1) that acquires each user's history information from the confidential history information storage means, and the similarity and recommendation degree are calculated from the acquired information. History information classifying means (for example, corresponding to the history information classifying unit 220 in FIG. 1) that extracts history information of a part of users including self or a part of history information associated with content included in the history of the user, Similarity information acquisition means for acquiring similarity information from the similarity storage means (for example, equivalent to the similarity information acquisition unit 240 in FIG. 1), and similarity calculation means for calculating similarity based on the classified history information (For example, equivalent to the second similarity calculation unit 230 in FIG. 1) and similarity estimation means for estimating similarity based on the calculated similarity and the acquired similarity information (for example, similarity estimation in FIG. 1) 250) and the recommended degree calculating means for calculating the recommended degree of content based on the estimated similarity, the calculated similarity, and the history information acquired by the history information classifying means (for example, the recommended degree shown in FIG. 1). A recommendation device characterized by comprising: a calculation unit 260) and a display means (for example, the display unit 270 in FIG. 1) for displaying content based on the calculated recommendation degree Yes.

  According to this invention, the disclosure history information storage means stores history information of a user who has permitted information disclosure such as purchase history and browsing history. On the other hand, the concealment history information storage means stores history information of users who are not permitted to be disclosed. The similarity storage means calculates similarity from the stored public history information, and stores the calculated similarity. The history information acquisition unit acquires the history information of each user from the history information storage unit. Here, the history information acquisition means acquires its own history information from the disclosure history information storage means for users who have permitted the disclosure, and from the confidential history information storage means for users who do not permit the disclosure. The history information classifying means extracts the history information of a part of users including the user who calculates the similarity from the acquired information or a part of the history information related to the content included in the history of the user. The similarity information acquisition unit acquires the similarity information from the similarity storage unit. The similarity calculation means calculates the similarity based on the acquired history information. The similarity estimation means estimates the similarity based on the calculated similarity and the acquired similarity information. The recommendation degree calculation means calculates the recommendation degree of the content based on the estimated similarity, the calculated similarity, and the history information acquired by the history information acquisition means. The display means displays the content based on the calculated recommendation level. Therefore, for users who are active in providing information, the similarity with other users is calculated based on the provided information, and the similarity with high accuracy obtained from the similarity information acquisition means and the acquired history information are calculated. Based on this, the recommendation level of the content is calculated. Therefore, a highly accurate recommendation result can be provided. For users who are reluctant to provide information, other users based on a part of the user's history information stored in the public history information storage means and their own history information stored in the confidential history information storage means. The similarity with other remaining users who have not calculated the similarity is estimated from the similarity obtained by the similarity information acquisition means, and the above calculation and the estimated similarity The recommendation level of the content is calculated based on the characteristics and the acquired history information. Therefore, a recommendation result can be provided even for a user who is reluctant to provide information.

  (2) The present invention defines similarity between users based on user history information such as purchase history and browsing history, and recommends content that is predicted to be of interest to the user based on the defined similarity. A recommendation device that uses the collaborative filtering method to output as a result, and that is a release history information storage means for storing the history information of the user who has permitted the release of the purchase history, browsing history, etc. (for example, the release history information database of FIG. 7) 110), a secret history information storage unit (for example, equivalent to the secret history information database 290 in FIG. 1) for storing history information of users who are not permitted to be disclosed, and similarity calculation from the stored history information. The similarity storage means for storing the calculated similarity (for example, equivalent to the similarity database 130 in FIG. 7) and the history information storage means. Is a first history information acquisition unit (for example, corresponding to the history information acquisition unit 211 of FIG. 7) that acquires each user's history information from the confidential history information storage unit, and calculates similarity from the acquired information History information classifying means (for example, corresponding to the history information classifying unit 220 in FIG. 7) that extracts history information of a part of users including self or a part of history information associated with content included in the history of the user, Similarity information acquisition means (for example, equivalent to the similarity information acquisition unit 240 in FIG. 7) for acquiring similarity information from the similarity storage means, and similarity calculation means for calculating similarity based on the acquired history information ( For example, it corresponds to the second similarity calculation unit 230 in FIG. 7) and similarity estimation means for estimating the similarity based on the calculated similarity and the acquired similarity information (for example, the similarity estimation unit 250 in FIG. 7). Phase ) And second history information acquisition means (for example, the history of FIG. 7) that acquires only necessary history information from the history information storage means based on the similarity obtained by the similarity calculation means and the similarity estimation means. The content recommendation degree based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated value of similarity acquired by the second history information acquisition unit. Recommendation level calculation means (for example, equivalent to the recommendation level calculation unit 260 in FIG. 7), and display means for displaying content based on the calculated recommendation level (for example, corresponding to the display unit 270 in FIG. 7). The recommendation apparatus characterized by having provided these.

  According to this invention, the disclosure history information storage means stores history information of a user who has permitted information disclosure such as purchase history and browsing history. On the other hand, the concealment history information storage means stores history information of users who are not permitted to be disclosed. The similarity storage means calculates similarity from the stored public history information, and stores the calculated similarity. The first history information acquisition unit acquires the history information of each user from the history information storage unit. Here, the first history information acquisition means acquires its own history information from the disclosure history information storage means for users permitted to be released and from the confidential history information storage means for users not permitted to be released. The history information classifying means extracts the history information of a part of users including the user who calculates the similarity from the acquired information or a part of the history information related to the content included in the history of the user. The similarity information acquisition unit acquires the similarity information from the similarity storage unit. The similarity calculation means calculates the similarity based on the acquired history information. The similarity estimation means estimates the similarity based on the calculated similarity and the acquired similarity information. The second history information acquisition unit acquires only necessary history information from the public history information storage unit or the secret history information storage unit based on the similarity obtained by the similarity calculation unit and the similarity estimation unit. The recommendation degree calculating means calculates the recommended degree of content based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated similarity value acquired by the second history information acquiring means. The display means displays the content based on the calculated recommendation level. Therefore, for users who are active in providing information, based on the provided information, calculate the similarity with other users, further obtain only the history information necessary to calculate the recommendation degree, The recommendation level of the content is calculated based on the calculated similarity and the history information of the acquired user having high similarity. Therefore, a highly accurate recommendation result can be provided. Also, for users who are reluctant to provide information, other users based on a part of the user's history information stored in the public history information storage means and their own history information stored in the confidential history information storage means. To calculate the degree of recommendation and to estimate the similarity with the other users who have not calculated the similarity based on the similarity obtained by the similarity information acquisition unit. Only the necessary history information is acquired, and the recommendation degree of the content is calculated based on the calculated similarity and the history information of the user having a high estimated value of the acquired similarity. Therefore, a highly accurate recommendation result can be provided even for a user who is reluctant to provide information. In addition, since the history information is retrieved each time before the recommendation level is calculated, the recommendation process can be performed even in an electronic device having a small memory capacity.

  (3) The present invention proposes a recommendation device characterized in that, for the recommendation device of (2), the second history information acquisition means selects and acquires history information of an upper user with high similarity. doing.

  According to this invention, the second history information acquisition unit selects and acquires the history information of the higher-order user with high similarity. Therefore, since the recommendation level is calculated based on the history information of the upper users having high similarity, it is possible to provide a recommendation result with higher accuracy.

  (4) The present invention proposes a portable terminal in which the recommendation device according to any one of claims 1 to 3 is built.

  According to this invention, since the recommendation device is built in the portable terminal, the user's history information can be easily acquired. Further, by calculating the similarity in advance, the processing load in the recommendation process can be reduced.

  (5) The present invention proposes a set top box (STB) in which the recommendation device according to any one of claims 1 to 3 is incorporated.

  According to the present invention, since the recommendation device is built in the set top box (STB), it is possible to easily acquire the user's history information. Further, by calculating the similarity in advance, the processing load in the recommendation process can be reduced.

  (6) The present invention relates to a history information database (for example, corresponding to the disclosure history information database 111 in FIG. 9) that stores user history information such as purchase history and browsing history for users who are permitted to disclose data. A data server (for example, equivalent to the similarity database 131 in FIG. 9) that defines and stores similarity between users based on user history information such as the purchase history and browsing history. , Corresponding to the server 100 in FIG. 9) and a recommendation device (for example, the recommendation device in FIG. 9) that uses a collaborative filtering method that outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result. 200), in which the recommendation device stores history information of itself. Storage means (for example, equivalent to the user history information storage unit 291 in FIG. 9) and first history information acquisition means (for example, the history information acquisition unit 210 in FIG. 9) that acquires its own history information from the history information storage means. And a second history information acquisition unit (e.g., equivalent to the history information acquisition unit 210 in FIG. 9) that acquires the history information of each user from the history information storage unit, and similar to the acquired information History information classifying means for extracting history information of a part of users including self for calculating the sex and recommendation degree, or a part of history information related to content included in the history of the user (for example, history information classification of FIG. 9) Unit 220), similarity information acquisition means for acquiring similarity information from the similarity database (for example, equivalent to the similarity information acquisition unit 240 of FIG. 9), and history information classification means Similarity calculation means for calculating similarity based on history information (for example, equivalent to the second similarity calculation unit 230 in FIG. 9), and similarity for estimating similarity based on the calculated similarity and acquired similarity information Based on the estimation means (for example, equivalent to the similarity estimation unit 250 in FIG. 9), the estimated similarity, the calculated similarity, and the history information acquired from the history information classification means, the recommendation level of the content is calculated. Recommendation level calculating means (for example, equivalent to the recommendation level calculating unit 260 in FIG. 9) and display means for displaying content based on the calculated recommendation level (for example, corresponding to the display unit 270 in FIG. 9). We propose a recommendation system characterized by this.

  According to the present invention, the history information storage means of the recommendation device stores its own history information. The first history information acquisition means acquires its own history information from the history information storage means. The second history information acquisition unit acquires the history information of each user from the history information storage unit. The history information classifying means calculates similarity and recommendation from the acquired information, and includes history information of some users including the user or a part of history information associated with content included in the user's history. Take out. The similarity information acquisition unit acquires similarity information from the similarity database. The similarity calculation unit calculates the similarity based on the history information acquired from the history information classification unit. The similarity estimation means estimates the similarity based on the calculated similarity and the acquired similarity information. The recommendation level calculation means calculates the recommendation level of content based on the estimated similarity, the calculated similarity, and the acquired history information. The display means displays the content based on the calculated recommendation level. Therefore, for users who are actively providing information, the similarity with other users is calculated based on the provided information, and the highly accurate similarity and history information classification means obtained from the similarity information acquisition means is used. The recommendation level of content is calculated based on the acquired history information. Therefore, a highly accurate recommendation result can be provided. For users who are reluctant to provide information, the similarity with other users is calculated based on a part of the information stored in the history information database and the information stored in the history information storage means. Based on the similarity obtained by the sex information acquisition means, the similarity with the remaining other users who have not calculated the similarity is inferred, and content recommendation based on the above calculated similarity and the acquired history information Calculate the degree. Therefore, a recommendation result can be provided even for a user who is reluctant to provide information.

  (7) The present invention relates to a history information database (for example, corresponding to the publication history information database 111 in FIG. 10) that stores user history information such as purchase history and browsing history for users who are permitted to publish data; A data server (for example, corresponding to the similarity database 131 in FIG. 10) that defines and stores similarity between users based on user history information such as purchase history and browsing history , Corresponding to the server 100 in FIG. 10) and a recommendation device using a collaborative filtering method that outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result (for example, the recommendation device in FIG. 10) The recommendation device stores its own history information. History information storage means (for example, equivalent to the user history information storage unit 291 in FIG. 10) and first history information acquisition means for acquiring own history information from the history information storage means (for example, history information acquisition in FIG. 10) Unit 211), a second history information acquisition unit (e.g., equivalent to the history information acquisition unit 211 of FIG. 10) that acquires each user's history information from the history information storage unit, and the acquired information History information classifying means (for example, the history information classifying unit of FIG. 10) that extracts history information of a part of users including self calculating similarity from the history information or a part of history information associated with content included in the history of the user 220), similarity information acquisition means for acquiring similarity information from the similarity database (for example, equivalent to the similarity information acquisition unit 240 in FIG. 10), and history information classification means Similarity calculation means (for example, corresponding to the second similarity calculation unit 230 in FIG. 10) that calculates similarity based on the obtained history information, and the similarity is estimated based on the calculated similarity and the acquired similarity information. From the similarity information obtained from the similarity estimation means (for example, equivalent to the similarity estimation unit 250 in FIG. 10) and the similarity obtained by the similarity calculation means and the similarity estimation means, only necessary history information is obtained from the history information database. Third history information acquisition means (for example, equivalent to the history information acquisition unit 211 in FIG. 10) to be acquired, the estimated similarity, the calculated similarity, and the similarity acquired from the third history information acquisition means A recommendation level calculation means (for example, equivalent to the recommendation level calculation unit 260 in FIG. 10) that calculates the recommendation level of the content based on the history information of the user having a high estimated value, and based on the calculated recommendation level A recommendation system characterized by comprising display means for displaying content (for example, corresponding to the display unit 270 in FIG. 10) is proposed.

  According to the present invention, the history information storage means of the recommendation device stores its own history information. The first history information acquisition means acquires its own history information from the history information storage means. The second history information acquisition unit acquires the history information of each user from the history information storage unit. The history information classifying means extracts the history information of some users including the user who calculates the similarity from the acquired information or a part of the history information associated with the content included in the user's history. The similarity information acquisition unit acquires similarity information from the similarity database. The similarity calculation unit calculates the similarity based on the history information acquired from the history information classification unit. The similarity estimation means estimates the similarity based on the calculated similarity and the acquired similarity information. The third history information acquisition unit acquires only necessary history information from the history information database based on the similarity obtained by the similarity calculation unit and the similarity estimation unit. The recommendation degree calculating means calculates the recommended degree of content based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated similarity value acquired from the third history information acquiring means. The display means displays the content based on the calculated recommendation level. Therefore, for users who are active in providing information, based on the provided information, calculate the similarity with other users, further obtain only the history information necessary to calculate the recommendation degree, The recommendation level of the content is calculated based on the calculated similarity and the history information of the acquired user having high similarity. Therefore, a highly accurate recommendation result can be provided. Also, users who are reluctant to provide information are similar to other users based on a part of the user's history information stored in the history information database and their own history information stored in the history information storage means. Necessary to calculate the degree of recommendation and to estimate the degree of recommendation based on the degree of similarity obtained by the similarity information acquisition means, inferring the similarity with the other users who have not calculated the similarity. Only the historical information is acquired, and the recommendation degree of the content is calculated based on the calculated similarity and the historical information of the user having a high estimated value of the acquired similarity. Therefore, a highly accurate recommendation result can be provided even for a user who is reluctant to provide information. In addition, since the history information is retrieved each time before the recommendation level is calculated, the recommendation process can be performed even in an electronic device having a small memory capacity.

  (8) The present invention proposes a recommendation system according to (7), wherein the third history information acquisition unit selects and acquires history information of a higher-order user with high similarity. doing.

  According to this invention, the third history information acquisition unit selects and acquires the history information of the higher-order user with high similarity. Therefore, since the recommendation level is calculated based on the history information of the upper users having high similarity, it is possible to provide a recommendation result with higher accuracy.

  (9) The present invention proposes a recommendation system in which the recommendation device is a portable terminal with respect to the recommendation system of (6) to (8).

  According to this invention, since the recommendation device is built in the portable terminal, the user's history information can be easily acquired. In addition, by calculating the similarity in advance on the server side, the processing load in the recommendation process can be reduced.

  (10) The present invention proposes a recommendation system in which the recommendation device is a set-top box (STB) for the recommendation system of (6) to (8).

  According to the present invention, since the recommendation device is built in the set top box (STB), it is possible to easily acquire the user's history information. In addition, by calculating the similarity in advance on the server side, the processing load in the recommendation process can be reduced.

  (11) The present invention defines similarity between users based on user history information such as purchase history and browsing history, and recommends content predicted to be of interest to the user based on the defined similarity. A recommendation method in a recommendation device using a collaborative filtering method that is output as a result, wherein the recommendation device stores history information of a user permitted to publish the purchase history, browsing history, etc. , Corresponding to the public history information database 110 in FIG. 1), secret history information storage means (for example, corresponding to the secret history information database 290 in FIG. 1) for storing history information of users who are not permitted to be released, and the stored Similarity storage means for calculating similarity from the public history information and storing the calculated similarity (for example, similarity data in FIG. 1) A first step (e.g., corresponding to step S101 in FIG. 6), and obtaining the history information of each user from the public history information storage means or the secret history information storage means A second step (for example, extracting a part of history information of a part of users including self calculating similarity and recommendation degree or a part of history information associated with content included in the history of the self (for example, Similar to step S102 of FIG. 6), a third step of obtaining similarity information from the similarity storage means (for example, equivalent to step S103 of FIG. 6), and calculating the similarity based on the classified history information A fourth step (e.g., corresponding to step S104 of FIG. 6) and a fifth step (example) of estimating similarity based on the calculated similarity and the acquired similarity information 6) (corresponding to step S105 in FIG. 6) and a sixth step (for example, calculating the recommendation degree of content based on the estimated similarity, the calculated similarity, and the history information acquired in the first step (for example, , Corresponding to step S106 in FIG. 6) and a seventh step (for example, corresponding to step S107 in FIG. 6) for displaying content based on the calculated recommendation degree Has proposed.

  According to this invention, the history information of a part of users including self that obtains the history information of each user from the public history information storage means or the secret history information storage means and calculates the similarity and the recommendation degree from the acquired information. Information or a part of history information associated with the content included in the own history is extracted. Next, the similarity information is acquired from the similarity storage means, the similarity is calculated from the classified history information, and the similarity is estimated from the calculated similarity and the acquired similarity information. Then, the recommendation level of the content is calculated based on the estimated similarity, the calculated similarity, and the history information acquired in the first step, and the content is displayed based on the calculated recommendation level. Therefore, for users who are actively providing information, the similarity with other users is calculated based on the provided information, and based on the highly accurate similarity obtained by the third step and the acquired history information. The content recommendation level is calculated. Therefore, a highly accurate recommendation result can be provided. For users who are reluctant to provide information, other users based on a part of the user's history information stored in the public history information storage means and their own history information stored in the confidential history information storage means. From the similarity obtained in the third step, the similarity with other remaining users who have not performed similarity calculation is estimated, the above calculation and the estimated similarity and The recommendation level of content is calculated based on the acquired history information. Therefore, a recommendation result can be provided even for a user who is reluctant to provide information.

  (12) The present invention defines similarity between users based on user history information such as purchase history and browsing history, and recommends content predicted to be of interest to the user based on the defined similarity. A recommendation method in a recommendation device that uses a collaborative filtering method that is output as a result, wherein the purchase history, the browsing history, etc. are published history information storage means for storing history information of users who have been permitted to publish (for example, the publication of FIG. 7 Similar to the history information database 110), secret history information storage means for storing history information of users who are not permitted to be released (for example, equivalent to the secret history information database 290 of FIG. 1), and similar to the stored history information Similarity storage means (for example, corresponding to the similarity database 130 of FIG. 7) for calculating the similarity and storing the calculated similarity; A first step of acquiring each user's history information from the history information storage means or the secret history information storage means (e.g., corresponding to step S201 in FIG. 8), and calculating similarity from the acquired information A second step (for example, corresponding to step S202 of FIG. 8) of extracting history information of a part of users including self or a part of history information associated with content included in the history of the user, and the similarity A third step (for example, corresponding to step S203 in FIG. 8) for acquiring similarity information from the storage means, and a fourth step (for example, in step S204 in FIG. 8) for calculating similarity based on the acquired history information. Equivalent), a fifth step (for example, corresponding to step S205 in FIG. 8) of estimating similarity based on the calculated similarity and the acquired similarity information, A sixth step (for example, corresponding to step S206 in FIG. 8) for obtaining only necessary history information from the history information storage means based on the similarity obtained in the step 4 and the fifth step, and the estimation A seventh step (for example, step of FIG. 8) that calculates the recommendation degree of the content based on the similarity information calculated, the calculated similarity, and the history information of the user having a high similarity estimated value acquired in the sixth step. And a recommendation method characterized by comprising an eighth step (for example, equivalent to step S208 in FIG. 8) for displaying content based on the calculated recommendation degree. .

  According to the present invention, the history information of each user is acquired from the history information storage means or the confidential history information storage means, and the history information of some users including the self calculating similarity from the acquired information, or A part of history information associated with the content included in the own history is extracted. Next, the similarity information is acquired from the similarity storage means, the similarity is calculated from the acquired history information, and the similarity is estimated from the calculated similarity and the acquired similarity information. Then, from the similarities obtained in the fourth step and the fifth step, only necessary history information is acquired from the history information storage means, and the estimated similarity, the calculated similarity, and the sixth step are acquired. The recommendation level of the content is calculated based on the history information of the user having a high similarity estimation value, and the content is displayed based on the calculated recommendation level. Therefore, for users who are active in providing information, based on the provided information, calculate the similarity with other users, further obtain only the history information necessary to calculate the recommendation degree, The recommendation level of the content is calculated based on the calculated similarity and the history information of the acquired user having high similarity. Therefore, a highly accurate recommendation result can be provided. Also, for users who are reluctant to provide information, other users based on a part of the user's history information stored in the public history information storage means and their own history information stored in the confidential history information storage means. In order to calculate the degree of recommendation from the similarity obtained in the third step, the similarity with other users who have not performed similarity calculation is estimated, and the degree of recommendation is calculated. Only the necessary history information is acquired, and the recommendation degree of the content is calculated based on the calculated similarity and the history information of the user having a high estimated value of the acquired similarity. Therefore, a highly accurate recommendation result can be provided even for a user who is reluctant to provide information. In addition, since the history information is retrieved each time before the recommendation level is calculated, the recommendation process can be performed even in an electronic device having a small memory capacity.

  (13) The present invention defines similarity between users based on user history information such as purchase history and browsing history, and recommends content that is predicted to be of interest to the user based on the defined similarity. A program for causing a computer to execute a recommendation method in a recommendation device that uses a collaborative filtering method that is output as a result, wherein the recommendation device stores history information of a user permitted to publish the purchase history, browsing history, etc. Public history information storage means (for example, corresponding to the public history information database 110 in FIG. 1) and confidential history information storage means (for example, the confidential history information database 290 in FIG. And the similarity is calculated from the stored public history information, and the calculated similarity is stored. A first step (for example, FIG. 1) of acquiring history information of each user from the public history information storage unit or the secret history information storage unit. 6), and the history information of a part of users including the user who calculates the similarity and the recommendation degree from the acquired information or the history information associated with the content included in the user's history. A second step (for example, corresponding to step S102 in FIG. 6), a third step for acquiring similarity information from the similarity storage means (for example, corresponding to step S103 in FIG. 6), A fourth step (for example, corresponding to step S104 in FIG. 6) for calculating similarity based on the classified history information, the calculated similarity, and the acquired similarity information Based on the fifth step (for example, corresponding to step S105 in FIG. 6), the estimated similarity, the calculated similarity, and the history information acquired in the first step. A sixth step (for example, corresponding to step S106 in FIG. 6) for calculating the recommendation degree, and a seventh step (for example, corresponding to step S107 in FIG. 6) for displaying content based on the calculated recommendation degree; , A program for causing a computer to execute is proposed.

  According to this invention, the history information of a part of users including self that obtains the history information of each user from the public history information storage means or the secret history information storage means and calculates the similarity and the recommendation degree from the acquired information. Information or a part of history information associated with the content included in the own history is extracted. Next, the similarity information is acquired from the similarity storage means, the similarity is calculated from the classified history information, and the similarity is estimated from the calculated similarity and the acquired similarity information. Then, the recommendation level of the content is calculated based on the estimated similarity, the calculated similarity, and the history information acquired in the first step, and the content is displayed based on the calculated recommendation level. Therefore, for users who are actively providing information, the similarity with other users is calculated based on the provided information, and based on the highly accurate similarity obtained by the third step and the acquired history information. The content recommendation level is calculated. Therefore, a highly accurate recommendation result can be provided. For users who are reluctant to provide information, other users based on a part of the user's history information stored in the public history information storage means and their own history information stored in the confidential history information storage means. From the similarity obtained in the third step, the similarity with other remaining users who have not performed similarity calculation is estimated, the above calculation and the estimated similarity and The recommendation level of content is calculated based on the acquired history information. Therefore, a recommendation result can be provided even for a user who is reluctant to provide information.

  (14) The present invention defines similarity between users based on user history information such as purchase history and browsing history, and recommends content predicted to be of interest to the user based on the defined similarity. A program for causing a computer to execute a recommendation method in a recommendation device that uses a collaborative filtering method to be output as a result, and storing history information of a user who is permitted to publish about the purchase history, browsing history, etc. Means (for example, equivalent to the disclosure history information database 110 in FIG. 7), confidential history information storage means (for example, equivalent to the confidential history information database 290 in FIG. 1) for storing history information of users who are not permitted to be disclosed, Similarity storage means for calculating similarity from stored public history information and storing the calculated similarity (for example, A first step (for example, corresponding to step S201 in FIG. 8) of acquiring history information of each user from the history information storage means or the confidential history information storage means, A second step of extracting history information of a part of users including the self for calculating similarity from the acquired information or a part of history information associated with content included in the history of the user (for example, FIG. 8 corresponding to step S202 of FIG. 8), a third step of acquiring similarity information from the similarity storage means (for example, corresponding to step S203 of FIG. 8), and a step of calculating similarity based on the acquired history information. 4 step (for example, corresponding to step S204 in FIG. 8) and a fifth step (for example, estimating similarity based on the calculated similarity and the acquired similarity information) , Corresponding to step S205 in FIG. 8) and a sixth step (for example, acquiring only necessary history information from the history information storage means based on the similarity obtained in the fourth step and the fifth step) , Corresponding to step S206 in FIG. 8), and the recommended degree of content based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated value of similarity acquired in the sixth step. A seventh step of calculating (for example, corresponding to step S207 in FIG. 8) and an eighth step of displaying content based on the calculated recommendation degree (for example, corresponding to step S208 in FIG. 8). Has proposed a program to be executed.

  According to the present invention, the history information of each user is acquired from the history information storage means or the confidential history information storage means, and the history information of some users including the self calculating similarity from the acquired information, or A part of history information associated with the content included in the own history is extracted. Next, the similarity information is acquired from the similarity storage means, the similarity is calculated from the acquired history information, and the similarity is estimated from the calculated similarity and the acquired similarity information. Then, from the similarities obtained in the fourth step and the fifth step, only necessary history information is acquired from the history information storage means, and the estimated similarity, the calculated similarity, and the sixth step are acquired. The recommendation level of the content is calculated based on the history information of the user having a high similarity estimation value, and the content is displayed based on the calculated recommendation level. Therefore, for users who are active in providing information, based on the provided information, calculate the similarity with other users, further obtain only the history information necessary to calculate the recommendation degree, The recommendation level of the content is calculated based on the calculated similarity and the history information of the acquired user having high similarity. Therefore, a highly accurate recommendation result can be provided. Also, for users who are reluctant to provide information, other users based on a part of the user's history information stored in the public history information storage means and their own history information stored in the confidential history information storage means. In order to calculate the degree of recommendation from the similarity obtained in the third step, the similarity with other users who have not performed similarity calculation is estimated, and the degree of recommendation is calculated. Only the necessary history information is acquired, and the recommendation degree of the content is calculated based on the calculated similarity and the history information of the user having a high estimated value of the acquired similarity. Therefore, a highly accurate recommendation result can be provided even for a user who is reluctant to provide information. In addition, since the history information is retrieved each time before the recommendation level is calculated, the recommendation process can be performed even in an electronic device having a small memory capacity.

  According to the present invention, a highly accurate recommendation result is output to a user who is active in providing information, and a recommendation result is also obtained from a user's history information recorded in the recommendation device for a user who is reluctant to provide information. Is effective. In addition, since the similarity calculation is performed by batch processing, and the recommendation device performs only the recommendation processing, there is an effect that it can be realized even in an electronic device with low processing capability.

It is a figure which shows the structure of the recommendation apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the estimation method of the similarity in the recommendation apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the estimation method of the similarity in the recommendation apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the estimation method of the similarity in the recommendation apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the estimation method of the similarity in the recommendation apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the process of the recommendation apparatus which concerns on the 1st Embodiment of this invention. It is a figure which shows the structure of the recommendation apparatus which concerns on the 2nd Embodiment of this invention. It is a figure which shows the process of the recommendation apparatus which concerns on the 2nd Embodiment of this invention. It is a figure which shows the structure of the recommendation system which concerns on the 3rd Embodiment of this invention. It is a figure which shows the structure of the recommendation system which concerns on the 4th Embodiment of this invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Note that the constituent elements in the present embodiment can be appropriately replaced with existing constituent elements and the like, and various variations including combinations with other existing constituent elements are possible. Therefore, the description of the present embodiment does not limit the contents of the invention described in the claims.

<First Embodiment>
A first embodiment of the present invention will be described with reference to FIGS.

<Configuration of recommendation device>
The configuration of the recommendation device according to the present embodiment will be described with reference to FIG. Note that the recommendation device according to the present embodiment is built in a portable terminal, an STB, or the like.

  As shown in FIG. 1, the recommendation device according to the present embodiment includes a public history information database 110, a first similarity calculation unit 120, a similarity database 130, a history information acquisition unit 210, and a history information classification unit. 220, a second similarity calculation unit 230, a similarity information acquisition unit 240, a similarity estimation unit 250, a recommendation degree calculation unit 260, a display unit 270, and a confidential history information database 290. Yes.

  The public history information database 110 stores the history information of users who have been permitted to publish, such as purchase history and browsing history, and anyone can use the stored history information for recommendations. Further, the confidential history information database 290 stores history information of users who are not permitted to be disclosed, and the stored history information can be used for recommendations only by the person himself / herself. The first similarity calculation unit 120 calculates the similarity between users based on the history information stored in the public history information database 110 and stores the calculation result in the similarity database 130.

  The history information acquisition unit 210 acquires the history information of each user from the public history information database 110 or the confidential history information database 290. The history information classification unit 220 obtains history information of a part of users including the user who calculates similarity and recommendation degree from the acquired information or a part of history information related to contents included in the history of the user. Take out. The second similarity calculation unit 230 calculates the similarity based on the classified history information. The similarity information acquisition unit 240 acquires similarity information from the similarity database 130.

  The similarity estimation unit 250 estimates the similarity based on the calculated similarity and the acquired similarity information. Details of the process will be described later. The recommendation level calculation unit 260 calculates the recommendation level of content based on the estimated similarity, the calculated similarity, and the history information acquired by the history information classification unit 220. Specifically, the recommendation degree of each content is determined based on the output result of the second similarity calculation unit 230 or the similarity that is the output result of the similarity estimation unit 250 and the history information that is the output result of the history information acquisition unit 210. For example, the similarity of each user and the presence / absence of the history of the content correspond to {1, 0}, and the multiplication sum is defined as the recommendation degree. The display unit 270 displays content based on the calculated recommendation level.

<Processing of similarity estimation unit>
The processing of the similarity estimation unit will be described with reference to FIGS.

  As shown in FIG. 2, a description will be given by taking user A to user D as an example. Here, the user A to the user D are users who have released the history information, and the first similarity calculation unit 120 calculates each other's similarity with a full mesh by batch processing, and calculates the calculation result. It is recorded in the similarity database 130. FIG. 3 is an example of data stored in the similarity database 130.

  The recommendation device performs a recommendation process for the user X who has not disclosed the history information. If the relationship between the user X and the user A to the user D is as shown in FIG. 4, the second similarity calculation unit 230 directly calculates the similarity for the user X, the user A, and the user B. In addition, since the user who has disclosed his / her history information has already obtained the similarity necessary for the recommendation in the similarity database 130, it is not necessary to perform this process and the following similarity estimation.

Next, as illustrated in FIG. 5, when the similarity calculation results of the user X, the user A, and the user B by the second similarity calculation unit 230 are “0.3” and “0.7”, respectively. The similarity between user A and user C (“0.5”), the similarity between user B and user C (“0.7”), and the similarity between user A and user D (“0.6”) Using the similarity (“0.4”) between the user B and the user D, for example, the similarity between the user X, the user C, and the user D is estimated by the following arithmetic expression.
Estimated value of similarity between user X and user C = (0.3 * 0.5 + 0.7 * 0.7) / 2
= 0.32
Estimated value of similarity between user X and user C = (0.3 * 0.6 + 0.7 * 0.4) / 2
= 0.23

<Processing of recommendation device>
The process of the recommendation apparatus according to the present embodiment will be described with reference to FIG.

  First, the history information of each user is acquired from the public history information database 110 or the confidential history information database 290 (step S101), and the history of some users including self calculating similarity and recommendation from the acquired information. Information or a part of history information associated with the content included in its own history is extracted (step S102). Next, similarity information is acquired from the similarity storage means (step S103), and the similarity is calculated from the classified history information (step S104).

  Then, the similarity is estimated based on the calculated similarity and the acquired similarity information (step S105), and the recommendation degree of the content is calculated based on the estimated similarity, the calculated similarity, and the history information acquired in step S101. (Step S106), and the content is displayed based on the calculated recommendation degree (Step S107).

  Therefore, according to the present embodiment, for users who are actively providing information, based on the provided information, the similarity with other users is calculated, and based on the calculated similarity and the acquired history information. The content recommendation level is calculated. Therefore, a highly accurate recommendation result can be provided. For users who are reluctant to provide information, based on the user's history information stored in the history information storage means, the similarity with other users is calculated, the estimated similarity, the calculated similarity and The recommendation level of content is calculated based on the acquired history information. Therefore, a recommendation result can be provided even for a user who is reluctant to provide information.

<Second Embodiment>
A second embodiment of the present invention will be described with reference to FIGS.

<Configuration of recommendation device>
The configuration of the recommendation device according to the present embodiment will be described with reference to FIG. Note that the recommendation device according to the present embodiment is built in a portable terminal, an STB, or the like.

  As shown in FIG. 7, the recommendation device according to the present embodiment includes a public history information database 110, a first similarity calculation unit 120, a similarity database 130, a history information acquisition unit 211, and a history information classification unit. 220, a second similarity calculation unit 230, a similarity information acquisition unit 240, a similarity estimation unit 250, a recommendation degree calculation unit 260, a display unit 270, a notification unit 280, and a confidential history information database 290. It consists of and. In addition, about the component which attaches | subjects the same code | symbol as 1st Embodiment, since it has the same function, the detailed description is abbreviate | omitted.

  The notification unit 280 sends a notification signal to the history information acquisition unit 211 so as to acquire only necessary history information from the public history information database 110 based on the similarity obtained by the second similarity calculation unit 230 and the similarity estimation unit 250. Send. The history information acquisition unit 211 acquires only necessary history information specified by the notification unit 280 from the public history information database 110.

<Processing of recommendation device>
Processing of the recommendation device according to the present embodiment will be described with reference to FIG.

  First, history information of each user is acquired from the public history information database 110 or the confidential history information database 290 (step S201), and history information of some users including self calculating similarity from the acquired information, or Then, a part of the history information associated with the content included in the own history is extracted (step S202). Next, the similarity information is acquired from the similarity storage means (step S203), and the similarity is calculated from the acquired history information (step S204).

  Then, the similarity is inferred from the calculated similarity and the acquired similarity information (step S205), and only necessary history information is acquired from the public history information database 110 based on the similarity obtained in steps S204 and S205. (Step S206).

  Then, the recommendation level of the content is calculated based on the estimated similarity, the calculated similarity, and the history information of the user having a high similarity estimated value acquired in step S206 (step S207), and based on the calculated recommendation level The content is displayed (step S208).

  Therefore, according to the present embodiment, for users who are actively providing information, the similarity with other users is calculated based on the provided information, and the recommendation level is further determined based on the estimated result. Only the history information necessary for the calculation is acquired, and the recommendation degree of the content is calculated based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated value of the acquired similarity. Therefore, a highly accurate recommendation result can be provided. Also, for users who are reluctant to provide information, the similarity with other users is calculated based on the user's history information stored in the history information storage means, and further recommended based on the estimated result. Only the history information necessary for calculating the degree is acquired, and the recommendation degree of the content is calculated based on the estimated similarity, the calculated similarity, and the history information of the user who has a high estimated value of the acquired similarity. Therefore, a highly accurate recommendation result can be provided even for a user who is reluctant to provide information. In addition, since the history information is retrieved each time before the recommendation level is calculated, the recommendation process can be performed even in an electronic device having a small memory capacity.

<Third Embodiment>
A third embodiment of the present invention will be described with reference to FIG. Note that the recommendation system according to the present embodiment includes a server 100 that provides history information necessary for a recommendation to a user recommendation device, and a recommendation device that performs a recommendation process based on the provided history information. In addition, the recommendation device according to the present embodiment is built in a mobile terminal, an STB, or the like.

<Configuration of recommendation system>
The configuration of the recommendation system according to this embodiment will be described with reference to FIG.

  As shown in FIG. 9, the recommendation system according to this embodiment includes a server 100 and a recommendation device 200. The server 100 includes a public history information database 111, a first similarity calculation unit 121, and a similarity database 131. The recommendation device 200 includes a history information acquisition unit 210, a history information classification unit 220, and the like. The second similarity calculation unit 230, the similarity information acquisition unit 240, the similarity estimation unit 250, the recommendation degree calculation unit 260, the display unit 270, and the user history information storage unit 291. . In addition, about the component which attaches | subjects the same code | symbol as 1st Embodiment, since it has the same function, the detailed description is abbreviate | omitted.

  The public history information database 111 stores history information of each user except the user himself. The first similarity calculation unit 121 calculates the similarity between users based on the history information stored in the public history information database 110 and stores the calculation result in the similarity database 131. The user history information storage unit 291 stores user history information.

  In the present embodiment, the server and the recommendation device share the processing that the recommendation device has collectively performed in the first embodiment and the second embodiment. For this reason, the processing load of the recommendation device is reduced and the memory capacity can be reduced. Therefore, the recommendation device can be built in an electronic device having a low processing capacity and a small memory capacity. For users who are actively providing information, the recommendation level of content is calculated based on the similarity information provided from the server and the history information acquired from the history information acquisition unit. Therefore, it is possible to provide a recommendation with high accuracy and a low processing load in the recommendation device. On the other hand, for users who are reluctant to provide information, the similarity with other users is calculated based on the information stored in the user history information storage unit, the estimated similarity, the calculated similarity, and the acquired history The recommendation level of content is calculated based on the information. Therefore, a recommendation result can be provided even for a user who is reluctant to provide information.

<Fourth Embodiment>
A fourth embodiment of the present invention will be described with reference to FIG. Note that the recommendation system according to the present embodiment includes a server 100 that provides history information necessary for a recommendation to a user recommendation device, and a recommendation device that performs a recommendation process based on the provided history information. In addition, the recommendation device according to the present embodiment is built in a mobile terminal, an STB, or the like.

<Configuration of recommendation system>
The configuration of the recommendation system according to the present embodiment will be described with reference to FIG.

  The recommendation system according to the present embodiment includes a server 100 and a recommendation device 200, as shown in FIG. The server 100 includes a public history information database 111, a first similarity calculation unit 121, and a similarity database 131. The recommendation device 200 includes a history information acquisition unit 210, a history information classification unit 220, and the like. From the second similarity calculation unit 230, the similarity information acquisition unit 240, the similarity estimation unit 250, the recommendation degree calculation unit 260, the display unit 270, the notification unit 280, and the user history information storage unit 291. It is configured. In addition, about the component which attaches | subjects the same code | symbol as 3rd Embodiment, since it has the same function, the detailed description is abbreviate | omitted.

  The recommendation system according to the present embodiment is obtained by adding the function in the second embodiment to the recommendation device in the recommendation system in the third embodiment. That is, in the first embodiment and the second embodiment, the server and the recommendation device share the processing that the recommendation device has collectively performed. For this reason, the processing load of the recommendation device is reduced and the memory capacity can be reduced. Therefore, the recommendation device can be built in an electronic device having a low processing capacity and a small memory capacity. For users who are actively providing information, based on the similarity information provided by the server, obtain only the history information necessary to calculate the recommendation level, and obtain the similarity and the obtained similarity. The recommendation level of the content is calculated based on the history information of the user having a high estimated value of sex. Therefore, it is possible to provide a recommendation with higher accuracy and less processing load in the recommendation device. Also, for users who are reluctant to provide information, the similarity with other users is calculated based on the information stored in the history information storage means, and the recommendation level is further calculated based on the estimated result. Therefore, only the history information necessary for this is acquired, and the recommendation degree of the content is calculated based on the estimated similarity, the calculated similarity, and the history information of the user who has a high estimated value of the acquired similarity. Therefore, a highly accurate recommendation result can be provided even for a user who is reluctant to provide information.

  Note that the recommendation device or the recommendation system of the present invention is realized by recording the processing of the recommendation device or the recommendation system on a computer-readable recording medium, causing the recommendation device to read and execute the program recorded on the recording medium. be able to. The computer system here includes an OS and hardware such as peripheral devices.

  Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW (World Wide Web) system is used. The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.

  The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

  The embodiments of the present invention have been described in detail with reference to the drawings. However, the specific configuration is not limited to the embodiments, and includes designs and the like that do not depart from the gist of the present invention. For example, in the present invention, the example in which the recommendation process is performed by paying attention to the similarity between users has been described, but the method shown here can also be used for items and the like.

  In the present invention, the history information to be used is acquired from a public database or a storage medium such as a hard disk or a memory inside the terminal using the history information acquisition unit and the similarity data acquisition unit. Assuming, but not necessarily limited to this.

  Further, the method for providing recommendation results according to the present invention assumes that a list describing the degree of recommendation of each content for the user is provided to an external recommendation presentation method. It is not limited to.

  In addition, as the history information used in the history information acquisition unit, a complete database is recorded that records the purchase history of products and the browsing history of web pages, but the database is prepared separately, without using the present invention. It can also be constructed by history information provided by a user who may disclose data.

  Further, from all the data in the database, for example, it is assumed that reduced data is created by selecting, for example, selecting the top several users with a large number of histories, and this is acquired by the history information acquisition unit. However, it is not limited to this method.

  Further, in the database, a vector in which the presence / absence of the history of each content is associated with {1,0} for all users is created, the similarity between each user is defined as a cosine coefficient between the vectors, and the numerical value is A recorded database may be prepared and acquired from the database by the similarity data acquisition unit. However, the similarity between users is not limited to this definition, and the timing of creating similarity data is not limited to this.

100; server 110; disclosure history information database 111; disclosure history information database 120; first similarity calculation unit 121; first similarity calculation unit 130; similarity database 131; similarity database 200; recommendation device 210; Information acquisition unit 211; History information acquisition unit 220; History information classification unit 230; Second similarity calculation unit 240; Similarity information acquisition unit 250; Similarity estimation unit 260; Recommendation degree calculation unit 270; Display unit 280; Unit 290; confidential history information database 291; user history information storage unit

Claims (14)

A collaborative filtering method that defines similarity between users based on user history information such as purchase history and browsing history, and outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result A recommendation device using
About the purchase history, browsing history, etc., a public history information storage means for storing history information of a user permitted to publish,
Concealment history information storage means for storing history information of users who are not permitted to be published,
Similarity storage means for calculating similarity from the stored publication history information and storing the calculated similarity;
History information acquisition means for acquiring history information of each user from the public history information storage means or confidential history information storage means;
History information classifying means for extracting history information of a part of users including the user who calculates similarity and recommendation from the acquired information or a part of history information associated with content included in the history of the user; ,
Similarity acquisition means for acquiring similarity from the similarity storage means;
A similarity calculation means for calculating similarity by the history information Eject by the history information classification means,
Similarity estimation means for estimating similarity based on the calculated similarity and the acquired similarity ;
Recommendation degree calculating means for calculating a recommendation degree of content based on the estimated similarity, the calculated similarity, and history information acquired by the history information classification means;
Display means for displaying content based on the calculated recommendation degree;
A recommendation device characterized by comprising: A collaborative filtering method that defines similarity between users based on user history information such as purchase history and browsing history, and outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result A recommendation device using
About the purchase history, browsing history, etc., a public history information storage means for storing history information of a user permitted to publish,
Concealment history information storage means for storing history information of users who are not permitted to publish,
Similarity storage means for calculating similarity from the stored publication history information and storing the calculated similarity;
First history information acquisition means for acquiring history information of each user from the public history information storage means or the confidential history information storage means;
History information classifying means for extracting a part of history information of a part of users including self calculating similarity from the acquired information or a part of history information related to contents included in the history of the user,
Similarity acquisition means for acquiring similarity from the similarity storage means;
Similarity calculating means for calculating similarity based on the acquired history information;
Similarity estimation means for estimating similarity based on the calculated similarity and the acquired similarity ;
Second history information acquisition means for acquiring only necessary history information from the history information storage means based on the similarity obtained by the similarity calculation means and the similarity estimation means;
A recommendation degree calculating means for calculating a recommendation degree of content based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated value of similarity acquired in the second history information acquiring means;
Display means for displaying content based on the calculated recommendation degree;
A recommendation device characterized by comprising:   The recommendation apparatus according to claim 2, wherein the second history information acquisition unit selects and acquires history information of an upper user having high similarity.   A portable terminal in which the recommendation device according to claim 1 is incorporated.   A set top box (STB) in which the recommendation device according to claim 1 is incorporated. For users who are permitted to publish data, the similarity between users based on the history information database storing user history information such as purchase history and browsing history and the user history information such as purchase history and browsing history A data server having a similarity database to be defined and stored, and a recommendation device using a collaborative filtering method that outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result A recommendation system,
The recommendation device has history information storage means for storing its own history information;
First history information acquisition means for acquiring own history information from the history information storage means;
Second history information acquisition means for acquiring history information of each user from the history information storage means;
History information classifying means for extracting history information of a part of users including the user who calculates similarity and recommendation from the acquired information or a part of history information associated with content included in the history of the user; ,
Similarity acquisition means for acquiring similarity from the similarity database;
Similarity calculating means for calculating similarity based on history information acquired from the history information classification means;
Similarity estimation means for estimating similarity based on the calculated similarity and the acquired similarity ;
Recommendation degree calculating means for calculating a recommendation degree of content based on the estimated similarity, the calculated similarity, and history information acquired from the history information classification means;
Display means for displaying content based on the calculated recommendation degree;
The recommendation system characterized by having. For users who are permitted to publish data, the similarity between users based on the history information database storing user history information such as purchase history and browsing history and the user history information such as purchase history and browsing history A data server having a similarity database to be defined and stored, and a recommendation device using a collaborative filtering method that outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result A recommendation system,
The recommendation device has history information storage means for storing its own history information;
First history information acquisition means for acquiring own history information from the history information storage means;
Second history information acquisition means for acquiring history information of each user from the history information storage means;
History information classifying means for extracting a part of history information of a part of users including self calculating similarity from the acquired information or a part of history information related to contents included in the history of the user,
Similarity acquisition means for acquiring similarity from the similarity database;
Similarity calculating means for calculating similarity based on history information acquired from the history information classification means;
Similarity estimation means for estimating similarity based on the calculated similarity and the acquired similarity ;
From the similarity obtained by the similarity calculation means and the similarity estimation means, third history information acquisition means for acquiring only necessary history information from the history information database;
A recommendation degree calculating means for calculating a recommendation degree of content based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated similarity value acquired from the third history information acquiring means;
Display means for displaying content based on the calculated recommendation degree;
The recommendation system characterized by having.   The recommendation system according to claim 7, wherein the third history information acquisition unit selects and acquires history information of an upper user having high similarity.   The recommendation system according to claim 6, wherein the recommendation device is a mobile terminal.   The recommendation system according to claim 6, wherein the recommendation device is a set top box (STB). A collaborative filtering method that defines similarity between users based on user history information such as purchase history and browsing history, and outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result A recommendation method in a recommendation device using
The recommendation device is, for the purchase history, browsing history, etc., public history information storage means for storing history information of users who are permitted to publish, secret history information storage means for storing history information of users who are not permitted to publish, Similarity storage means for calculating similarity from the stored publication history information and storing the calculated similarity;
A first step of acquiring history information of each user from the public history information storage means or the secret history information storage means;
A second step of extracting, from the acquired information, history information of a part of users including the user who calculates the similarity and recommendation degree, or a part of history information associated with the content included in the history of the user; ,
A third step of obtaining similarity from the similarity storage means;
A fourth step of calculating the similarity by taking out the history information in the second step,
A fifth step of inferring similarity based on the calculated similarity and the acquired similarity ;
A sixth step of calculating a recommendation level of the content based on the estimated similarity, the calculated similarity, and the history information acquired in the first step;
A seventh step of displaying content based on the calculated recommendation degree;
A recommendation method characterized by comprising: A collaborative filtering method that defines similarity between users based on user history information such as purchase history and browsing history, and outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result A recommendation method in a recommendation device using
With regard to the purchase history, browsing history, etc., public history information storage means for storing history information of users who are permitted to publish, confidential history information storage means for storing history information of users who are not permitted to publish, and the stored disclosure Similarity storage means for calculating the similarity from the history information and storing the calculated similarity;
A first step of acquiring history information of each user from the history information storage means or the confidential history information storage means;
A second step of extracting history information of a part of users including self calculating similarity from the acquired information or a part of history information associated with content included in the history of the user;
A third step of obtaining similarity from the similarity storage means;
A fourth step of calculating similarity based on the acquired history information;
A fifth step of inferring similarity based on the calculated similarity and the acquired similarity ;
From the similarity obtained in the fourth step and the fifth step, a sixth step of acquiring only necessary history information from the history information storage means,
A seventh step of calculating a recommendation level of content based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated value of similarity acquired in the sixth step;
An eighth step of displaying content based on the calculated recommendation degree;
A recommendation method characterized by comprising: A collaborative filtering method that defines similarity between users based on user history information such as purchase history and browsing history, and outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result A program for causing a computer to execute a recommendation method in a recommendation device using
The recommendation device is, for the purchase history, browsing history, etc., public history information storage means for storing history information of users who are permitted to publish, secret history information storage means for storing history information of users who are not permitted to publish, Similarity storage means for calculating similarity from the stored publication history information and storing the calculated similarity;
A first step of acquiring history information of each user from the public history information storage means or the secret history information storage means;
A second step of extracting, from the acquired information, history information of a part of users including the user who calculates the similarity and recommendation degree, or a part of history information associated with the content included in the history of the user; ,
A third step of obtaining similarity from the similarity storage means;
A fourth step of calculating the similarity by taking out the history information in the second step,
A fifth step of inferring similarity based on the calculated similarity and the acquired similarity ;
A sixth step of calculating a recommendation level of the content based on the estimated similarity, the calculated similarity, and the history information acquired in the first step;
A seventh step of displaying content based on the calculated recommendation degree;
A program that causes a computer to execute. A collaborative filtering method that defines similarity between users based on user history information such as purchase history and browsing history, and outputs content predicted to be of interest to the user based on the defined similarity as a recommendation result A program for causing a computer to execute a recommendation method in a recommendation device using
With regard to the purchase history, browsing history, etc., public history information storage means for storing history information of users who are permitted to publish, confidential history information storage means for storing history information of users who are not permitted to publish, and the stored disclosure Similarity storage means for calculating the similarity from the history information and storing the calculated similarity;
A first step of acquiring history information of each user from the history information storage means or the confidential history information storage means;
A second step of extracting history information of a part of users including self calculating similarity from the acquired information or a part of history information associated with content included in the history of the user;
A third step of obtaining similarity from the similarity storage means;
A fourth step of calculating similarity based on the acquired history information;
A fifth step of inferring similarity based on the calculated similarity and the acquired similarity ;
From the similarity obtained in the fourth step and the fifth step, a sixth step of acquiring only necessary history information from the history information storage means,
A seventh step of calculating a recommendation level of content based on the estimated similarity, the calculated similarity, and the history information of the user having a high estimated value of similarity acquired in the sixth step;
An eighth step of displaying content based on the calculated recommendation degree;
A program that causes a computer to execute.

Images