JP2012079223A - Recommendation coordination device and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a recommendation coordination device and its method capable of determining contents to be recommended to specific users by coordinating a user-based recommendation device and a content-based popularity stake rank recommendation device and adding information related to the popularity state rank for each content.SOLUTION: A recommendation coordination device 4 corrects priority stake score of a candidate list received by a candidate list reception part 41 based on popularity stake score received by a popularity stake rank reception part 42, by using a candidate list correction part 43. The recommendation coordination device 4 transmits the candidate list corrected by the candidate list correction part 43 as the corrected candidate list from a corrected candidate list transmission part 44 to the user-based recommendation device 2.

Description

  The present invention relates to a recommendation cooperation apparatus and a method thereof.

  In recent years, content recommendations that provide effective content to users who use Web sites such as an electronic commerce site have become widespread. For example, a method has been proposed in which the interest and preference of the specific user is estimated based on the access history of the specific user to the website, and content corresponding to the estimated information is distributed to the specific user (for example, a patent). Reference 1).

  On the other hand, based on the access history from an unspecified number of users for each content, the popularity of the content to the unspecified number of users is measured, and based on this popularity, the content is distributed to the unspecified number of users. A method has been proposed.

Japanese Patent Application Laid-Open No. 2004-94983

  By the way, the method of recommending content to a specific user and the method of recommending based on the popularity rank of the content differ in the timing of updating the target user and the data that is the basis of the recommendation. . In other words, in the former method, segmentation is performed based on user attribute information and behavior history, or recommendations are made by specifying individuals, whereas in the latter method, a plurality of users are identified without specifying users. In many cases, recommendations are made without distinction. On the other hand, with the former method, the frequency of updating is relatively low even though the data that serves as the basis for the recommendation is periodically updated, whereas with the latter method, the selection history of each user is almost real time. In order to reflect these, etc., data that is the basis of the recommendation is frequently updated.

  Therefore, in the former method, when a candidate content to be recommended is given, it is possible to immediately recommend appropriate content for each user, but it is difficult to desire improvement thereafter. . On the other hand, in the latter method, in order to make a recommendation reflecting each user's selection history, etc. in near real time, content that has been gaining popularity is recommended preferentially. While it is possible to improve the accuracy of the recommendation, immediately after the content candidate to be recommended is given, such selection history etc. has not been obtained, so it is only possible to select the content to recommend at random Therefore, it was not possible to make recommendations with high accuracy from the beginning.

  Therefore, the present invention can determine the content to be recommended to a specific user by linking the recommendation device for each user and the popularity rank recommendation device for each content and taking into account information related to the popularity rank for each content. It is an object of the present invention to provide a recommendation collaboration device and method thereof.

  (1) A recommendation cooperation device that links a user-specific recommendation device that performs content recommendation for each user and a content-specific popularity rank recommendation device that performs content recommendation based on the popularity rank of the content. In addition, the identification information of the content to be recommended to the user, the keyword corresponding to the content, and the priority score indicating the priority of the recommendation corresponding to the keyword are associated from the recommendation device for each user. Candidate list receiving means for receiving information as a candidate list, and content corresponding to content identification information contained in the candidate list received by the candidate list receiving means from the content popularity rank recommendation device Popularity rank receiving means for receiving a popularity score indicating the popularity degree of content as a popularity rank for each content, and the candidate list receiving means received based on the popularity score received by the popularity rank receiving means Candidate list correcting means for correcting the priority score of the candidate list; corrected candidate list transmitting means for transmitting the candidate list corrected by the candidate list correcting means to the recommendation device for each user as a corrected candidate list; Recommendation cooperation device comprising

  In the recommendation cooperation device of (1), the candidate list receiving unit corresponds to the identification information of the content to be recommended to the user from the recommendation device for each user, the keyword corresponding to the content, and the keyword. Information associated with a priority score indicating the priority of the recommendation is received as a candidate list, and is included in the candidate list received by the candidate list receiving unit from the popularity rank recommendation device for each content by the popularity rank receiving unit. A popularity score indicating the popularity of the content corresponding to the content identification information is received as the popularity rank by content. And the recommendation cooperation apparatus of (1) correct | amends the priority score of the candidate list received by the candidate list receiving means based on the popularity score received by the popularity rank receiving means by the candidate list correcting means, The corrected candidate list transmission means transmits the corrected candidate list as a corrected candidate list to the recommendation device for each user.

  Therefore, since the recommendation cooperation apparatus of (1) corrects the priority score of the content to be recommended to the user based on the popularity rank by content, information related to the popularity rank by content The content to be recommended to a specific user can be determined. That is, since the recommendation cooperation apparatus of (1) reflects the popularity of the content in the entire user at the present time, an improvement in the click-through rate of the content having a high overall popularity can be expected even in the recommendation for each user. Therefore, the recommendation cooperation device of (1) is popular by content that can be expected to further improve the click-through rate over time in addition to the effect of recommendation by user who can expect a high click-through rate from the beginning. The effect of degree rank recommendation can be superimposed.

  (2) Storage means for storing the popularity score received by the popularity rank receiving means over a predetermined period, and the candidate list correcting means includes the popularity score received by the popularity rank receiving means and the storage means. Based on the stored popularity score, the time series change of the popularity score is analyzed, and the priority score of the candidate list received by the candidate list receiving unit is corrected based on the analysis result. The recommendation cooperation apparatus as described in (1).

  The recommendation cooperation apparatus of (2) corrects the priority score based on the time series change of the popularity score. By doing in this way, since the recommendation cooperation apparatus of (2) can reflect the time-sequential change of a popularity score on a priority score, for example, a plurality of contents having the same popularity rank at the present time In the case where there is, the content in the rising phase of the popularity rank can be recommended in preference to the content in the falling phase of the popularity rank.

  (3) Recommendation that links a user-specific recommendation device that performs content recommendation for each user and a content-specific popularity rank recommendation device that performs content recommendation based on the popularity rank of the content for unspecified users This is a method executed by the cooperation device, wherein the identification information of the content to be recommended to the user from the recommendation device for each user, the keyword corresponding to the content, and the priority of the recommendation corresponding to the keyword Receiving the information associated with the priority score shown as a candidate list, and receiving the candidate list from the content-specific popularity rank recommendation device. Based on the popularity score received in the step of receiving the popularity score indicating the popularity of the content corresponding to the identification information of the content as the popularity rank by content, and the step of receiving the popularity rank by content, In the recommendation apparatus for each user, the candidate list corrected in the step of correcting the priority score of the candidate list received in the step of receiving the candidate list and the step of correcting the priority score is used as the corrected candidate list. Transmitting.

  According to such a method, the same effect as in (1) can be expected by performing the method.

  ADVANTAGE OF THE INVENTION According to this invention, the content recommended to a specific user can be determined in cooperation with the recommendation apparatus classified by user and the popularity rank recommendation apparatus classified by content, and considering the information concerning the popularity rank classified by content. A recommendation cooperation device and method thereof can be provided.

It is a figure which shows the function outline | summary of the recommendation cooperation system which concerns on this embodiment. It is a figure which shows user keyword DB which concerns on this embodiment. It is a figure which shows content keyword DB which concerns on this embodiment. It is a figure which shows popularity degree DB classified by content concerning this embodiment. It is a figure which shows the example which correct | amends a priority score based on the time series change of the popularity score which concerns on this embodiment. It is a flowchart which shows the flow of a process in the recommendation cooperation system which concerns on this embodiment. It is a figure which shows an example of the candidate list | wrist which concerns on this embodiment. It is a figure which shows an example of the corrected candidate list | wrist which concerns on this embodiment. It is a figure which shows the relationship between time and a click through rate at the time of recommending with the conventional recommendation apparatus. It is a figure which shows the relationship between time and a click through rate at the time of recommending with the recommendation apparatus classified by user using the recommendation cooperation apparatus which concerns on this embodiment.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[Function configuration]
FIG. 1 is a diagram showing a functional outline of the recommendation cooperation system 1 according to the present embodiment. The recommendation cooperation system 1 includes a user-specific recommendation device 2 that performs content recommendation for each user and a content-specific popularity rank recommendation device 3 that performs content recommendation based on the popularity rank of the content for unspecified users. And the recommendation cooperation apparatus 4 which cooperates the recommendation apparatus 2 classified by user and the popularity rank recommendation apparatus 3 classified by content, and the user terminal 5.

  The recommendation cooperation device 4 can be connected to the recommendation device 2 by user and the popularity rank recommendation device 3 by content via a communication network such as a LAN or a WAN. Further, the user terminal 5 can be connected to the recommendation device 2 by user and the popularity rank recommendation device 3 by content via a communication network.

  This embodiment is applied to a computer (user recommendation device 2, content popularity rank recommendation device 3, recommendation cooperation device 4, user terminal 5) and its peripheral devices. Each unit in the present embodiment is configured by hardware included in a computer and its peripheral devices, and software that controls the hardware.

  The hardware includes a CPU as a control unit, a storage unit, a communication unit, a display unit, and an input unit. Examples of the storage unit include a memory (RAM, ROM, etc.), a hard disk drive (HDD), and an optical disk (CD, DVD, etc.) drive. Examples of the communication unit include various wired and wireless interface devices. Examples of the display unit include various displays such as a liquid crystal display and a plasma display. Examples of the input unit include a keyboard and a pointing device (mouse, tracking ball, etc.).

  The software includes a computer program and data for controlling the hardware. The computer program and data are stored in the storage unit, and are appropriately executed and referenced by the control unit. The computer program and data can be distributed via a communication network, and can also be stored and distributed on a computer-readable medium such as a CD-ROM.

Next, the functional configuration of the user-specific recommendation device 2 will be described.
The control unit of the user-specific recommendation device 2 includes a page request reception unit 21, a keyword acquisition unit 22, a content matching unit 23, a content determination unit 24, and a page transmission unit 25. In addition, the storage unit of the user-specific recommendation device 2 includes a user keyword database (hereinafter referred to as a DB) 26, a content keyword DB 27, and a content DB 28 that stores information on content and Web pages.

  The page request receiving unit 21 receives a Web page transmission request from the user terminal 5 and receives a Web page acquisition request, and acquires a user ID indicating user identification information of the user terminal 5. Here, the user ID may be obtained by providing an authentication function, for example, or the terminal browser may be specified using a cookie.

  The keyword acquisition unit 22 acquires, from the user keyword DB 26, a user keyword as a keyword corresponding to the user ID acquired by the page request reception unit 21 in response to reception of a web page transmission request by the page request reception unit 21. To do.

  FIG. 2 is a diagram showing the user keyword DB 26 according to the present embodiment. The user keyword DB 26 stores a user ID, a user keyword corresponding to the user ID, and a user relevance score indicating the relevance of the user keyword with the user in association with each other.

  Here, the user keyword indicates the user's demographics and behavior attributes, and is given to the user ID based on, for example, the user's profile information, purchase history in the Web shop, and access history to the Web page. The The user relevance score indicates the degree of relevance between the user keyword and the user. For example, when the user has a relatively high access frequency with respect to a specific web page, the specific relevance score The user relevance score of the user keyword corresponding to the Web page is set high.

  FIG. 3 is a diagram showing the content keyword DB 27 according to the present embodiment. The content keyword DB 27 stores a content ID that identifies content, a content name, a content keyword that indicates a keyword related to the content, and a content relevance score that indicates the degree of relevance between the content of the content keyword and the content keyword. .

  The content matching unit 23 refers to the content keyword DB 27, specifies content corresponding to the content keyword that matches the user keyword acquired by the keyword acquisition unit 22, and generates a candidate list of content to be recommended. Subsequently, the content matching unit 23 transmits the generated candidate list to the recommendation cooperation device 4.

  The candidate list includes a content ID, a content name, a content keyword, and a priority score. The priority score is a score obtained based on the user relevance score corresponding to the user keyword and the content relevance score corresponding to the content keyword, and indicates the priority of the content recommendation. For example, the priority score is calculated by multiplying the user relevance score and the content relevance score. The priority score is calculated by multiplying the user relevance score and the content relevance score. However, the priority score is not limited to this and may be calculated by another method.

  The content determination unit 24 receives a corrected candidate list described later from the recommendation cooperation device 4. Then, the recommendation cooperation apparatus 4 determines the content added with respect to the web page transmitted to the user terminal 5 based on the corrected candidate list. For example, for each content included in the corrected list, the content determination unit 24 totals the priority scores of the content keywords corresponding to the content, and adds the content having the highest total value to the content to be added to the Web page. decide.

  The page transmission unit 25 acquires the content determined by the content determination unit 24 from the content DB 28 and embeds this content in the Web page for which the acquisition request has been made by the page request reception unit 21. Subsequently, the page transmission unit 25 transmits the Web page in which the content is embedded to the user terminal 5 that has requested transmission of the Web page.

Next, the functional configuration of the popularity rank recommendation device 3 by content will be described.
The control unit of the popularity rank recommendation device 3 by content includes a click-through data reflection unit 31. In addition, the storage unit of the content-specific popularity rank recommendation device 3 includes a content-specific popularity DB 32.

  FIG. 4 is a diagram showing the content popularity DB 32 according to the present embodiment. The content popularity DB 32 stores the content ID and the popularity score indicating the popularity of the content corresponding to the content ID in association with each other as the popularity rank for each content.

  The click-through data reflecting unit 31 transmits content data in a timely manner to the user terminal 5 and receives log data (click-through log data) of content that has received a selection request from the user terminal 5 (clicked by the user). In other words, the click-through data reflection unit 31 acquires click-through log data from the user terminal 5 and updates the content popularity DB 32. Specifically, the click-through data reflection unit 31 corrects the popularity score of the content ID corresponding to the acquired click-through log data so as to be relatively higher than other popularity scores. The click-through log data is data indicating that the user terminal 5 has transitioned to a page such as an advertisement or content when a link such as an advertisement or content is clicked.

Next, the functional configuration of the recommendation cooperation device 4 will be described.
The control unit of the recommendation cooperation device 4 includes a candidate list receiving unit 41 (candidate list receiving unit), a popularity rank receiving unit 42 (popularity rank receiving unit), and a candidate list correcting unit 43 (candidate list correcting unit). A corrected candidate list transmission unit 44 (corrected candidate list transmission means).

  The candidate list receiving unit 41 receives a candidate list from the user-specific recommendation device 2. Specifically, the candidate list receiving unit 41 receives the content ID of the content to be recommended to the user from the recommendation device 2 by user, the content name corresponding to the content ID, the content keyword corresponding to the content, And information associated with the priority score corresponding to the content keyword is received as a candidate list.

  The popularity rank receiving unit 42 receives the popularity score corresponding to the content ID included in the candidate list received by the candidate list receiving unit 41 as the popularity rank by content from the popularity rank recommendation device 3 by content. Specifically, the popularity rank receiving unit 42 transmits an acquisition request for the popularity rank by content to the popularity rank recommendation device 3 by content in response to receiving the candidate list by the candidate list reception unit 41. . Then, the popularity rank receiving unit 42 receives the popularity rank by content from the popularity rank recommendation device 3 by content.

  The candidate list correcting unit 43 corrects the priority score included in the candidate list received by the candidate list receiving unit 41 based on the popularity rank by content received by the popularity rank receiving unit 42. Specifically, the candidate list correction unit 43 makes the priority score corresponding to the content ID relatively higher than the other priority scores for the content ID having a relatively high popularity score. To correct. Further, the candidate list correcting unit 43 corrects the priority score corresponding to the content ID of the content ID having a relatively low popularity score so as to be relatively lower than the other priority scores. .

  In this embodiment, the candidate list correcting unit 43 corrects the priority score included in the candidate list received by the candidate list receiving unit 41 based on the popularity score received by the popularity rank receiving unit 42. However, it is not limited to this. For example, the popularity score received by the popularity rank receiving unit 42 is stored in the storage unit for a predetermined period, and the candidate list correcting unit 43 stores the latest popularity score received by the popularity rank receiving unit 42 and the memory. Based on the past popularity score stored in the section, the time series change of the popularity score is analyzed, and the priority score included in the candidate list received by the candidate list receiving section 41 is calculated based on the analysis result. It is good also as correcting. By doing in this way, since the recommendation cooperation apparatus 4 can correct | amend a priority score based on the time-sequential change of a popularity score, for example, the some content which has an equivalent popularity rank at the present time is In such a case, it is possible to recommend the content in the rising phase of the popularity rank in preference to the content in the falling phase of the popularity rank.

  FIG. 5 is a diagram illustrating an example in which the priority score is corrected based on the time series change of the popularity score according to the present embodiment. In FIG. 5, the horizontal axis indicates the date, and the vertical axis indicates the popularity score. In FIG. 5, the popularity score of “9/22” is the past popularity score stored in the storage unit, and the popularity score of “9/27” is the latest received by the popularity rank receiving unit 42. The popularity score. Since the content “Auction: Railway Model Special Feature” is in a phase of decreasing popularity, the candidate list correction unit 43 indicates a priority score for this content, showing a tendency for the popularity to be flat, The content is corrected to be lower than the content having the degree score. In addition, since the content “travel: sleeper express” is in a phase of increasing popularity, the candidate list correction unit 43 indicates that the priority score of this content tends to be flat, The content is corrected so as to be higher than the content having the popularity score.

  The corrected candidate list transmission unit 44 transmits the candidate list corrected by the candidate list correction unit 43 to the recommendation device 2 for each user as a corrected candidate list.

[flowchart]
Next, the process flow in the recommendation cooperation system 1 will be described.
FIG. 6 is a flowchart showing the flow of processing in the recommendation cooperation system 1 according to this embodiment.

In step S <b> 1, the control unit (page request receiving unit 21) of the user-specific recommendation device 2 receives a Web page transmission request and a user ID from the user terminal 5.
In step S2, the control unit (keyword acquisition unit 22) of the recommendation device 2 for each user refers to the user keyword DB 26, and acquires a user keyword corresponding to the user ID accepted in step S1.

  In step S3, the control unit (content matching unit 23) of the user-specific recommendation device 2 refers to the content keyword DB 27, identifies the content corresponding to the content keyword that matches the user keyword acquired in step S2, A candidate list of contents to be recommended is generated.

  FIG. 7 is a diagram illustrating an example of a candidate list according to the present embodiment. In the candidate list shown in FIG. 7, it can be confirmed that the content name, the content keyword, and the priority score are associated with the content ID. For example, in the candidate list shown in FIG. 7, it can be confirmed that the contents “Auction: Special feature on railway model” and “Travel: Sleeper limited express” are included.

  Returning to FIG. 6, in step S <b> 4, the control unit (content matching unit 23) of the user-specific recommendation device 2 transmits the candidate list generated in step S <b> 3 to the recommendation cooperation device 4.

In step S5, the control unit (candidate list reception unit 41) of the recommendation cooperation device 4 receives the candidate list transmitted in step S4.
In step S6, the control unit (popularity rank receiving unit 42) of the recommendation cooperation device 4 receives the candidate list in step S5, and sends the popularity rank by content to the popularity rank recommendation device 3 by content. Send rank acquisition request.

In step S7, the control unit of the content-specific popularity rank recommendation device 3 receives the content-specific popularity rank acquisition request transmitted in step S6.
In step S8, the control unit of the content popularity rank recommendation device 3 transmits the content popularity rank to the recommendation cooperation device 4 in response to receiving the content popularity rank acquisition request in step S7. To do.

  In step S9, the control part (popularity rank receiving part 42) of the recommendation cooperation apparatus 4 receives the popularity rank classified by content transmitted in step S8. Here, it is assumed that the popularity rank by content shown in FIG. 4 is received.

  In step S10, the control unit (candidate list correction unit 43) of the recommendation cooperation device 4 is included in the candidate list received in step S5 based on the popularity rank (popularity score) by content received in step S9. The candidate list is corrected by correcting the priority score. The corrected candidate list is a corrected candidate list.

  For example, in the popularity rank by content shown in FIG. 4, the content “Travel: Sleeping Express” corresponding to the content ID “10198” corresponds to the content “Auction: Railway Model Special Feature” corresponding to the content ID “14410”. Since the popularity score is relatively high, the priority score of the content keyword corresponding to “Travel: Sleeper Express” is corrected to be relatively high, and the priority of the content keyword corresponding to “Auction: Special feature on railway model” is corrected. The degree score is corrected to be relatively low.

  FIG. 8 is a diagram showing an example of the corrected candidate list according to the present embodiment. In the corrected candidate list shown in FIG. 8, it can be confirmed that the content name, the content keyword, and the priority score are associated with the content ID. Further, in the corrected candidate list, the priority score corresponding to the contents “auction: railway model special feature” and “travel: sleeper express” included in the candidate list shown in FIG. 7 is changed. I can confirm.

  Returning to FIG. 6, in step S <b> 11, the control unit (corrected candidate list transmission unit 44) of the recommendation cooperation device 4 sets the candidate list corrected in step S <b> 10 as the corrected candidate list to the user-specific recommendation device 2. Send.

In step S12, the control unit (content determination unit 24) of the user-specific recommendation device 2 receives the corrected candidate list transmitted in step S11.
In step S <b> 13, the control unit (content determination unit 24) of the user-specific recommendation device 2 determines the content to be added to the Web page transmitted to the user terminal 5 based on the corrected candidate list.

  In step S14, the control unit (page transmission unit 25) of the user-specific recommendation device 2 embeds the content determined in step S13 into the web page requested for acquisition in step S1, and embeds this web page. The data is transmitted to the user terminal 5 that has requested transmission of the Web page.

  Subsequently, the click-through rate (CTR: Click Through Rate) in the user-specific recommendation device 2 when the recommendation cooperation device 4 according to the present embodiment is used, the CTR in the conventional user-specific recommendation device, and the conventional The difference with the CTR in the popularity rank recommendation device classified by contents is explained.

  FIG. 9 is a diagram illustrating a relationship between time and a click-through rate when a recommendation is performed using a conventional recommendation device. Specifically, FIG. 9A is a diagram illustrating a relationship between time and a click-through rate when the recommendation is performed by the conventional popularity rank recommendation device by content. FIG.9 (b) is a figure which shows the relationship between time and a click through rate at the time of recommending with the conventional recommendation apparatus according to user.

In conventional content-specific popularity rank recommendation apparatus, as shown in FIG. 9 (a), the value of the click-through rate tends to gradually rise from an initial value V _1. In other words, in the conventional popularity rank recommendation device by content, in order to make a recommendation reflecting each user's selection history etc. in almost real time, it is recommended that the content that is increasing in popularity is preferentially recommended, etc. refinement of the recommendation is made in accordance with the course of, as a result, the value of the click-through rate tends to gradually rise from an initial value V _1.

In contrast, in a conventional user-specific recommendations device, as shown in FIG. 9 (b), the value of the click-through rate tends almost unchanged from the initial value V _2. In other words, in the conventional recommendation device for each user, appropriate content is immediately recommended for each user by performing segmentation based on the user's attribute information and action history, or by specifying an individual and making a recommendation. it is possible to Deshon, initial value V ₂ becomes a higher value than the initial value V ₁ in the conventional content-specific popularity rank recommendation device. However, in the conventional recommendation device for each user, since improvement in accuracy cannot be expected even if time elapses, there is a tendency that hardly changes from the initial value V2.

FIG. 10 is a diagram illustrating a relationship between time and a click-through rate when the recommendation cooperation device 4 according to the present embodiment is used and the recommendation is performed by the user-specific recommendation device 2. When using the recommendations linkage device 4 according to the present embodiment, the initial value of the click-through rate in the user-specific recommendations device V ₃ becomes, be higher than the initial value V ₂ in a conventional user-specific recommendations device I can confirm. Thereafter, it can be confirmed that further increases from the initial value V _3. In addition to the effect of user-specific recommendations that can be expected to have a high click-through rate from the beginning, the effect of popularity rank recommendation by content that can be expected to further improve the click-through rate over time is superimposed. Because it can.

  As described above, according to the present embodiment, the recommendation cooperation device 4 causes the candidate list receiving unit 41 to use the content ID to be recommended to the user from the recommendation device 2 for each user, the content keyword corresponding to the content, and Information associated with the priority score indicating the priority of the recommendation corresponding to the content keyword is received as a candidate list, and the popularity rank receiving unit 42 receives the candidate list from the popularity rank recommendation device 3 by content. The popularity score indicating the popularity of the content corresponding to the content ID included in the candidate list received by the unit 41 is received as the popularity rank by content. And the recommendation cooperation apparatus 4 correct | amends the priority score of the candidate list received by the candidate list receiving part 41 by the candidate list correction | amendment part 43 based on the popularity score received by the popularity rank receiving part 42, The corrected candidate list transmission unit 44 transmits the corrected candidate list as a corrected candidate list to the recommendation device 2 for each user.

  Therefore, since the recommendation cooperation apparatus 4 corrects the priority score of the content to be recommended to the user of the user terminal 5 based on the popularity rank by content, the recommendation cooperation device 4 relates to the popularity rank by content. In consideration of the information, the content to be recommended to the specific user can be determined. In other words, the recommendation cooperation device 4 reflects the popularity of the content in the entire current user in the recommendation for each user, and therefore, in the recommendation for each user, the click-through rate of the content that has a high popularity for the entire user. Improvement can be expected. Therefore, in addition to the effect of the recommendation for each user who can expect a high click-through rate from the beginning, the recommendation cooperation device 4 can further increase the click-through rate with the passage of time. The effect of the foundation can be superimposed.

  As mentioned above, although embodiment of this invention was described, this invention is not restricted to embodiment mentioned above. The effects described in the embodiments of the present invention are only the most preferable effects resulting from the present invention, and the effects of the present invention are limited to those described in the embodiments of the present invention. is not.

  In the present embodiment, the user-specific recommendation apparatus 2 performs the user-specific recommendation. However, the present invention is not limited to this, and the recommendation may be performed based on segments to which a plurality of users belong.

DESCRIPTION OF SYMBOLS 1 Recommendation cooperation system 2 Recommendation apparatus according to user 3 Popularity rank recommendation apparatus according to contents 4 Recommendation cooperation apparatus 5 User terminal 41 Candidate list receiving part (candidate list receiving means)
42 Popularity rank receiving part (Popularity rank receiving means)
43 Candidate list correction unit (priority score correction means)
44 corrected candidate list transmission unit (corrected candidate list transmission means)

Claims (3)

It is a recommendation cooperation device that links a recommendation device by user that performs content recommendation by user and a popularity rank recommendation device by content that performs content recommendation based on the popularity rank of the content,
Information associated with the identification information of the content to be recommended to the user, the keyword corresponding to the content, and the priority score indicating the priority of the recommendation corresponding to the keyword from the recommendation device for each user. Candidate list receiving means for receiving as a candidate list;
The popularity score indicating the popularity of the content corresponding to the content identification information included in the candidate list received by the candidate list receiving unit is received as the popularity rank by content from the content-specific popularity rank recommendation device. Popularity rank receiving means,
Candidate list correcting means for correcting the priority score of the candidate list received by the candidate list receiving means based on the popularity score received by the popularity rank receiving means;
A recommendation cooperating apparatus comprising: a corrected candidate list transmitting unit configured to transmit the candidate list corrected by the candidate list correcting unit as a corrected candidate list to the recommendation device for each user. Storage means for storing the popularity score received by the popularity rank receiving means over a predetermined period;
The candidate list correcting means analyzes a time series change of the popularity score based on the popularity score received by the popularity rank receiving means and the popularity score stored in the storage means, and the analysis The recommendation cooperation apparatus according to claim 1, wherein the priority score of the candidate list received by the candidate list receiving unit is corrected based on a result. In a method executed by a recommendation cooperation device that links a recommendation device by user that performs content recommendation by user and a popularity rank recommendation device by content that performs content recommendation based on the popularity rank of the content There,
Information associated with the identification information of the content to be recommended to the user, the keyword corresponding to the content, and the priority score indicating the priority of the recommendation corresponding to the keyword from the recommendation device for each user. Receiving as a candidate list;
The popularity score indicating the popularity of the content corresponding to the identification information of the content included in the candidate list received in the step of receiving the candidate list from the content-specific popularity rank recommendation device as the popularity rank by content Receiving step;
Correcting the priority score of the candidate list received in the step of receiving the candidate list based on the popularity score received in the step of receiving the popularity rank by content;
Transmitting the candidate list corrected in the step of correcting the priority score to the recommendation device for each user as a corrected candidate list.

Images