JP5275126B2 - RECOMMENDATION INFORMATION GENERATION DEVICE AND RECOMMENDATION INFORMATION GENERATION METHOD - Google Patents

Description

  The present invention relates to a recommendation information generation apparatus and a recommendation information generation method.

  The Content-basedFiltering method is considered as a method for recommending content related to the content that the user is currently viewing. In this method, when a user takes an action such as browsing a Web page, the browsed target document is analyzed, and words (keywords) included in the document are extracted. Then, it is considered to provide information to the user using these words. For example, a system that enables recommendation of an appropriate advertisement by matching a content currently viewed by a user with a keyword assigned to an advertisement product is generally known.

  As a document related to such a technique, the document of Patent Document 1 (Japanese Patent Laid-Open No. 2007-148878) is cited. This Patent Document 1 describes that a distribution destination for distributing recommendation information and recommendation information to be distributed are determined and distributed in accordance with a predetermined recommendation extraction rule.

JP 2007-148878 A

  However, in Content-based Filtering, it is possible to output content related to the content that the user is currently viewing, but only the same content can be presented to all users. Therefore, it is not possible to provide appropriate content according to the user.

  Therefore, an object of the present invention is to provide a recommendation information generation apparatus and a recommendation information generation method that can generate recommendation information in consideration of a user's browsing history and content currently being browsed.

  In order to solve the above-described problem, a recommendation information generation apparatus according to the present invention includes a browsing content acquisition unit that acquires content specified by a user, and a browsing that indicates history content that is the content acquired by the browsing content acquisition unit. Browsing history information storing means for storing history information separately for each user, a recommendation request receiving means for receiving a recommendation request including a content ID indicating the content currently being browsed, and received by the recommendation request receiving means History similarity calculation means for calculating the similarity between the content specified by the content ID included in the recommendation request and each history content stored in the browsing history information storage means, and the history similarity calculation means The history similarity calculated by, and the user's The history evaluation value calculation means for calculating the correction evaluation value of the history content by multiplying the value, and the history are rearranged in the descending order of the correction evaluation value calculated by the history evaluation value calculation means so that the arrangement order is satisfied. User feature vector calculation means for calculating a user feature vector, and calculating the similarity between the user feature vector of the user calculated by the user feature vector calculation means and the feature vector of the recommendation target content to be recommended, Recommended content determining means for determining recommendation information to be recommended content for the user.

  Further, the recommendation information generation method of the present invention includes a browsing content acquisition step for acquiring content specified by a user, and browsing history information indicating history content that is the content acquired by the browsing content acquisition step for each user. Included in the recommendation request received by the storing step for distinguishing and storing in the browsing history information storage means, the recommendation request receiving step for receiving the recommendation request including the content ID indicating the currently browsed content, and the recommendation request receiving step Calculated by the history similarity calculation step for calculating the similarity between the content identified by the content ID being stored and each history content stored in the browsing history information storage means, and the history similarity calculation step. History similarity and history content The history evaluation value calculation step for calculating the corrected evaluation value of the history content by multiplying the user's evaluation value for the history, and the history is rearranged in the descending order of the correction evaluation value calculated by the history evaluation value calculation step. A user feature vector calculation step for calculating a user feature vector so as to satisfy, and a similarity between the user feature vector calculated by the user feature vector calculation step and the feature vector of the recommendation target content to be recommended By doing this, there is provided a recommended content determination step for determining recommendation information to be recommended content for the user.

  According to this invention, first, content designated by a user is acquired, and browsing history information indicating history content that is the acquired content is stored separately for each user.

  When a recommendation request including a content ID indicating the content currently being browsed is received, the similarity between the content specified by the content ID and each history content already stored is calculated, and the calculated history is calculated. The history content correction evaluation value is calculated by multiplying the similarity by the user's evaluation value for the history content. Then, the history is rearranged in the descending order of the calculated correction evaluation value, the feature vector of the user is calculated so as to satisfy the arrangement order, and the calculated user feature vector of the user and the feature vector of the recommendation target content to be recommended The recommendation information that is the content to be recommended for the user is determined.

  This makes it possible to determine recommendation information that is content suitable for the user based on the content being browsed and the history content, improve the accuracy of the recommendation information, and provide the user with more appropriate content. be able to.

  Further, in the recommendation information generating apparatus of the present invention, the recommendation request receiving means receives a recommendation request including a user situation vector together with a content ID, and stores a situation vector storage for storing a user situation vector indicating a past user situation. And a situation indicated by a user situation vector indicating a current situation received by the recommendation request receiving means and a situation indicated by a past user situation vector stored in the situation vector storage means Situation similarity calculation means for calculating situation similarity, and the history evaluation value calculation means further multiplies the situation similarity calculated by the situation similarity calculation means with a user evaluation value for history content The multiplication value is calculated with the history similarity calculated by the history similarity calculating means and the history code. By adding the correction evaluation value obtained by multiplying the evaluation value of the user for Ceiling, it is preferable to calculate the correction evaluation value in consideration of the user status.

  According to the present invention, a recommendation request including a user situation vector together with a content ID is received, and the situation indicated by the received user situation vector indicating the current situation is close to the situation indicated by the past user situation vector. The situation similarity indicating the above is calculated, and the calculated value of the situation similarity is multiplied by the user's evaluation value for the history content. By adding the value to the correction evaluation value obtained by multiplying the value, a correction evaluation value in consideration of the user situation can be calculated. A user feature vector can be generated according to the corrected evaluation value, and recommendation information can be generated based on the user feature vector. Therefore, it is possible to generate appropriate recommendation information according to the current user situation.

  According to the present invention, it is possible to determine recommendation information that is content suitable for the user based on the content being browsed and the history content, it is possible to improve the accuracy of the recommendation information, and to provide more appropriate content to the user. Can be provided.

1 is a system configuration diagram showing the entire system of a mobile device and an information distribution server of the present embodiment. 3 is a block diagram illustrating functions of a mobile device 100 and an information distribution server 200. FIG. 2 is a hardware configuration diagram of an information distribution server 200. FIG. It is explanatory drawing which shows user log | history information. It is explanatory drawing which shows the management information of a content. It is explanatory drawing which shows the memory content of a content table. It is a flowchart which shows the process at the time of determining the recommendation information of the information delivery server 200. FIG. A content characteristic vector X _c, is a schematic view showing a process of calculating the cosine value with the content characteristic vector X _i. It is a schematic diagram which shows a user condition vector. It is a schematic view showing a process of calculating the cosine value with the content characteristic vector X _c and the content characteristic vector X _i. Similarity sim (Cs, S _n) is a schematic view showing a process in which the evaluation value R to have been multiplied. It is a schematic diagram which shows the process in which EV value is calculated. It is a schematic diagram which shows the result as which the rearrangement process was performed based on EV value. It is a schematic diagram which shows the process of calculating EV value when not using condition reflection mode. It is a schematic diagram which shows the result of having rearranged according to EV value when not using condition reflection mode. It is a schematic diagram which shows the process at the time of calculating a user feature vector using the history content rearranged based on EV value in the case of a situation reflection mode. It is a schematic diagram which shows the process in which the predicted value P is calculated for every delivery object content. It is a schematic diagram which shows the result by which the content used as each delivery object was rearranged according to the predicted value P. FIG.

  Embodiments of the present invention will be described with reference to the accompanying drawings. Where possible, the same parts are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a system configuration diagram showing the entire system of a mobile device and an information distribution server of this embodiment. As shown in FIG. 1, this system includes a mobile device 100 and an information distribution server 200. The mobile device 100 receives content distributed from the information distribution server 200 or the like and is distributed. Can be viewed on the mobile device 100.

  Next, configurations of the mobile device 100 and the information distribution server 200 will be described. FIG. 2 is a block diagram illustrating functions of the mobile device 100 and the information distribution server 200. As illustrated in FIG. 2, the mobile device 100 includes a history transmission unit 10, a recommendation request transmission unit 12, a situation estimation unit 14, and a recommendation information display unit 16. The information distribution server 200 also includes a situation similarity calculation unit 20 (situation similarity calculation unit), a history similarity calculation unit 21 (history similarity calculation unit), a history evaluation value calculation unit 22 (history evaluation value calculation unit), User feature vector calculation unit 24 (user feature vector calculation unit), content category evaluation value calculation unit 26 (recommended content determination unit), history collection unit 28 (browsing content acquisition unit), recommendation request reception unit 30 (recommendation request reception unit) , A content transmission unit 32, a user history management DB 40 (browsing history information storage means), a content management DB 42, and a user feature vector management DB 44.

  The mobile device 100 or the information distribution server 200 is realized by the hardware configuration shown in FIG. FIG. 3 is a hardware configuration diagram of the information distribution server 200. As shown in FIG. 3, the information distribution server 200 shown in FIG. 2 is physically a data transmission / reception device such as a CPU 201, a RAM 202 and a ROM 203, which are main storage devices, an output device 205 such as a display, and a network card. The computer system includes a communication module 206 and an auxiliary storage device 207 such as a hard disk. Each function described in FIG. 2 has the input device 204, the output device 205, and the communication module 206 under the control of the CPU 201 by loading predetermined computer software on the hardware such as the CPU 201 and the RAM 202 shown in FIG. This is realized by reading and writing data in the RAM 202 and the auxiliary storage device 207. Since the mobile device 100 has the same hardware configuration, the description thereof is omitted here. Hereinafter, each functional block will be described based on the functional blocks shown in FIG.

  The history transmission unit 10 is a part that transmits history information such as browsing history and access history in the mobile device 100 to the information distribution server 200. The history information includes a terminal ID that identifies the terminal, a content ID that identifies the content, an evaluation value of the content evaluated by the user, and situation information (user situation vector). These processes may be transmitted by a user operation or at a predetermined cycle, or may be transmitted according to a predetermined rule.

  The recommendation request transmission unit 12 is a part that transmits a distribution request for recommendation information to the information distribution server 200 in accordance with a user operation. The recomment request includes a terminal ID, a situation ID, browsing information indicating the content currently being browsed, current user status information (situation vector), content type information, and the like. This situation ID is flag information indicating a distribution request for recommendation information considering the user situation. This flag information can be specified by the user. The browsing information includes a content ID that identifies the content and an evaluation value of the content evaluated by the user.

  The situation estimation unit 14 is a part that estimates the situation where the mobile device 100, that is, the user is placed. For example, it is a part for estimating position information indicating where the user is, the current time, and the situation around the user (there is nobody or how many are there). The position information can be acquired by using a known position positioning technique such as GPS technique, and the current time can be acquired by a built-in timer. Further, the situation around the user can determine which user is around a certain user based on GPS (registered information) of the user (mobile terminal) registered in advance. Further, it may be possible to determine the presence or absence of a surrounding person by a temperature sensor, a vibration sensor, or the like, or to allow the user to input the surrounding situation by manual input. The situation information indicating the situation of the user estimated by the situation estimation unit 14 is output to the recommendation request transmission unit 12 and used at the time of the recommendation request.

  The recommendation information display unit 16 is a part that displays the recommendation information distributed from the information distribution server 200.

  Next, the information distribution server 200 will be described. The situation similarity calculation unit 20 is included in the user situation information (situation vector) included in the recommendation request received by the recommendation request receiving unit 30 and the user history information collected in the past. This is a part for comparing the situation information (situation vector) of the user stored in, and calculating the similarity indicating the proximity of the situation.

The situation similarity calculation unit 20 is included in the user situation vector CS indicating the situation of the user at the time of the recommendation request received by the recommendation request receiving unit 30 (that is, the current situation) and the history information stored in the user history management DB 40. This is a part for calculating the similarity sim (CS, S _n ) with the situation vector S _{u, n} .

Here, the user situation vector CS indicating the current user situation is expressed by the following equation. S ₀ , S ₁ ... S _S represents a vector element in each situation parameter. For example, S ₀ is a parameter indicating a situation such as “I am at home” and S ₁ is “I am at a company”.

Then, the situation similarity calculation unit 20 calculates the situation similarity sim (CS, S _n ) between the user situation vector CS and the user situation vector S _{u, n} using the following equation (2). .

History similarity calculating unit 21, by the user to calculate the content characteristic vector X _c of the specified content is the content currently being viewed, the cosine value with the content characteristic vector X _i in the history content, specify the content and history content It is a part which calculates the similarity for judging the proximity.

Here, the content feature vector _Xc of the content being browsed is expressed as (3) below.

Then, using the following expression (4), the similarity sim (X _c , X _i ) between the content feature vector X _c of the designated content being browsed and the content feature vector X _i of the history content is similar to the history. Calculated by the degree calculator 21. Here, i is an identification code (for example, content ID) indicating history content, and H is a set of history content.

なお、ＥＶ値が同値の場合は、履歴登録時間が現在時間に近い方の履歴を優先して並び替える。また、状況反映モードでない行動ターゲティング機能の場合においては、ｓｉｍ（Ｘ_ｃ，Ｘ_ｉ）と評価値Ｒとの乗算値からＥＶ値を求める。 The history evaluation value calculation unit 22 includes a situation similarity sim (CS, S _n ) indicating the similarity between the user situation vector CS and the user situation vector S _{u, n,} and history contents that are contents browsed by the user in the past. The evaluation value R associated with the predetermined weighting factor α, the similarity sim (X _c , X _i ) indicating the similarity to the history content, and the evaluation value R indicating the evaluation This is a part for calculating the EV value as the correction evaluation value by calculating the sum of the multiplication values with the predetermined weight coefficient β. In consideration of the case where the same content is evaluated a plurality of times, a plurality of EV values for the same content are added as in the following equation (5). Here, N is the total number of contents i evaluated. Calculation of the following equation is obtained for all contents included in the history, and rearrangement is performed in descending order of EV value.

When the EV values are the same value, the history with the history registration time closer to the current time is preferentially sorted. In the case of a behavioral targeting function that is not in the situation reflection mode, the EV value is obtained from the product of sim (X _c , X _i ) and the evaluation value R.

  The user feature vector calculation unit 24 rearranges the history contents in descending order of the corrected evaluation value EV value calculated by the history evaluation value calculation unit 22, and calculates a user feature vector so as to satisfy the arrangement order. Specifically, the user feature vector calculation unit 24 is a part that calculates a user feature vector based on the current user status and the specified content being browsed, and the priority calculated by the history evaluation value calculation unit 22 The user feature vector is calculated so as to satisfy the history content rearranged according to

［参考文献］ T. Joachims, Optimizing SearchEnginesUsing Clickthrough Data
”, Proc. ACM SIGKDD Int’l Conf. KnowledgeDiscoveryand Data Mining (KDD02), AC
M Press, pp.132-142, 2002.
In this embodiment, a user feature vector can be calculated using a technique called RankingSVM proposed by Joachims et al. This RankingSVM is a technique that can obtain a user's feature vector so as to satisfy the order relationship between the feature points given priority in the feature space. In RankingSVM, the following equation (6) is used. The user feature vector W can be obtained by calculating the user feature vector W that minimizes V (w, ξ) under the conditions. In the following equation (6), w represents a user feature vector, and x represents a content feature vector. Here, both the user feature vector and the content feature vector are made up of vector elements indicating the category type, for example, made up of vector elements of the category type shown in FIG.

[References] T. Joachims, Optimizing SearchEnginesUsing Clickthrough Data
”, Proc. ACM SIGKDD Int'l Conf. KnowledgeDiscovery and Data Mining (KDD02), AC
M Press, pp.132-142, 2002.

The content category evaluation value calculation unit 26 obtains an inner product of the user feature vector calculated by the user feature vector calculation unit 24 and each content feature vector in the history content, and uses the obtained inner product value to obtain an evaluation value for the content category. This is a part for calculating the corresponding predicted value P. The content category evaluation value calculation unit 26 calculates a predicted value (Prediction) P _{u, i} of the evaluation for the content i of the user u as shown in the following equation (7). W _u is a user feature vector of the user u, and X _i is a content feature vector of the content i.

  The history collection unit 28 is a part that collects history information (history content etc.) such as browsing history or access history transmitted from the mobile device 100 together with status information (user status vector) and stores it in the user history management DB 40. .

The recommendation request receiving unit 30 is a part that receives a recommendation request (including a user situation vector, a content ID, and the like) transmitted from the mobile device 100 (recommendation request transmitting unit 12). When the recommendation request receiving unit 30 receives a recommendation request from the mobile device 100, the recommendation request receiving unit 30 notifies the situation similarity calculating unit 20, the history similarity calculating unit 21, and the content transmitting unit 32 to that effect, and the content feature vector and the like are transmitted. The processing is executed so that the contents i are rearranged in accordance with the predicted value P _u, i of the evaluation of the user u for the contents i calculated by using the information and transmitted as recommendation information.

When the content transmission unit 32 receives a rearrangement / recommendation information transmission instruction from the recommendation request reception unit 30, the content transmission unit 32 transmits the content list rearranged according to the predicted value _{Pu, i} as the recommendation information as described above.

  The user history management DB 40 is a database that stores user history information indicating a user's browsing history or access history. FIG. 4 is an explanatory diagram showing user history information. As shown in FIG. 4, the user history information includes date / time information indicating the date of browsing or access, identification information for identifying the content browsed or accessed, an evaluation value R of the content, and a user's place when browsing or accessing the user history information. The situation information (user situation vector) indicating the current situation is described in association with each other. The evaluation value R may be a numerical value input by the user, or may be a numerical value calculated based on a predetermined rule, for example, a rule that increases the evaluation value when the content browsing time is long.

  The situation information (user situation vector) is acquired based on the history information collected by the history collection unit 28 described above. As shown in FIG. 4, the status information is composed of three items: place, time, and people who are together, and the place is indicated by 1 or 0 indicating whether it is a home or a company. In addition, the time is indicated by 1 or 0 in the morning or night. The number of people who are together is 1 or 0, indicating whether they are alone, with their colleagues, or with their family.

  Here, the situation information is defined based on a predetermined criterion, but the present invention is not limited to this. For example, the positions of the mobile devices 100 are periodically collected and clustered. , It may be divided into k groups, and the situation information may be defined based on the divided groups. Specifically, if clustering is performed to divide into groups such as position a and position b, history information indicating what kind of content was viewed when in which group is generated. Also good.

  Further, the user history management DB 40 may store history content and its content feature vector. Specifically, the user history management DB 40 associates browsing date and time, content ID for specifying content, user evaluation, and content feature vector. As described later, the content feature vector is stored in the content management DB 42, and calculation processing related to history content is performed using the content feature vector stored in the DB. Also good.

  The content management DB 42 is a database that stores management information of each content. FIG. 5 is an explanatory diagram showing content management information. As shown in FIG. 5, a content ID, which is identification information for specifying content, and each feature vector are stored in association with each other. Here, items such as sports, news, variety, and music are allocated as feature vector items, and the feature vector is configured so that “1” is related and “0” is not related.

  The user feature vector management DB 44 is a part that stores calculated user feature vectors.

  The content table 46 is a database that stores specific content information. FIG. 6 is an explanatory diagram showing the contents stored in the content table. As shown in FIG. 6, a recommendation text to be distributed is stored in association with each other as content ID, content category, content title, and recommended distribution information. The content table 46 is not an essential configuration, and necessary content may be acquired from an external server.

  Next, the operation of the information distribution server 200 configured as described above will be described. FIG. 7 is a flowchart showing a process when determining recommendation information of the information distribution server 200.

  In the information distribution server 200, the history collection unit 28 periodically collects history information and stores the collected history information in the user history management DB 40. On the other hand, when a recommendation request is made when the mobile device 100 acquires and browses one content, the information distribution server 200 receives the recommendation request, and user status information included in the recommendation request. The browsing information is acquired, and the current user status information and the browsing information regarding the currently browsed content are temporarily stored in a predetermined storage unit (S101).

Then, the history similarity calculation unit 21 calculates a cosine value between the content feature vector X _c of the designated content currently being browsed and the content feature vector X _i of the history content stored in the user history management DB 40. (S102). This calculation example will be described with reference to FIG. Figure 8 is a schematic view showing a process of calculating the content characteristic vector X _c of the specified content browsing, the cosine value with the content characteristic vector X _i in the history content. As is apparent from FIG. 8, the history similarity calculation unit 21 calculates the similarity represented by the cosine value for each history content.

Then, depending on whether or not a situation request includes a situation ID, the recommendation request receiving unit 30 determines whether or not the situation reflection mode is set (S103). If it is determined that the recommendation request receiving unit 30 is in the situation reflection mode, the user similarity information of the current user is acquired by the situation similarity calculation unit 20 (S104), and the user is the past user situation information. The situation similarity sim (CS, S _n ) between the situation vector S _n and the user situation vector CS which is the situation information of the current user is calculated (S105).

This calculation example will be described with reference to FIGS. FIG. 9A is a schematic diagram showing a user status vector CS indicating the current user status information, and FIG. 9B is a user status vector S _n indicating the status of the user who has browsed past history content. It is a schematic diagram which shows. As is clear from FIG. 9, the user situation vector is composed of elements representing places such as homes and companies, elements representing surrounding situations such as one person or many people, and elements representing time zones such as morning, noon, and night. Yes.

And this calculation example is demonstrated using FIG. Figure 10 is a schematic diagram showing the current user situation vector CS, status similarity sim (CS, S _n) of the user state vector S _n in the history content process of calculating. As shown in FIG. 10, the situation similarity between the current user situation vector and the user situation vector in the history content is calculated.

  FIG. 11 shows a process in which the similarity calculated here is multiplied by the evaluation value. As shown in FIG. 11, the similarity calculated for each content is multiplied by an evaluation value.

The product of the situation similarity sim (CS, S _n ), the evaluation value R, and the weighting factor α, and the multiplication value of the history similarity sim (X _c , X _i ), the evaluation value R, and the weighting factor β. The corrected evaluation value EV value which is the sum is calculated by the history evaluation value calculation unit 22. Based on this corrected evaluation value EV value, the history contents are rearranged by the user feature vector calculation unit 24 (S106).

Specific examples thereof will be described with reference to FIGS. As shown in FIG. 12, the product of the situation similarity sim (CS, S _n ), the evaluation value R, and the weighting factor α (= 0.5), and the browsing content similarity sim (X _c , X _i ), The evaluation value R, and the multiplication value of the weighting coefficient β (= 0.5) are added to calculate an EV value for each content.

  Further, as shown in FIG. 13, the contents are rearranged based on the EV value. As shown in FIGS. 12 and 13, since the user browses the content ID “I0006” twice, the value is calculated as an EV value by adding the two according to the above formula. Accordingly, the ranking based on the EV value is also rearranged in the higher order by the content ID 0006 "corresponding to the EV value.

  Then, the user feature vector calculation unit 24 calculates the user feature vector W so as to satisfy the arrangement order (S107).

  FIG. 16 shows a specific example. FIG. 16 is a schematic diagram illustrating a process when calculating user feature vectors using history content rearranged based on EV values in the case of the situation reflection mode. As shown in FIG. 16, user feature vectors are calculated using the SVM method so as to satisfy the order of the history content.

Next, a predicted value P indicating the similarity between the user feature vector W calculated here and the content feature vector X _i of the content to be distributed is calculated by the content category evaluation value calculation unit 26 (S108). The content is rearranged by the content transmission unit 32 according to the predicted value P, and recommendation information is generated (S109). Then, the recommendation information is distributed by the content transmission unit 32 (S10).

  The processing here will be described with reference to FIGS. As shown in FIG. 17, a predicted value P indicating the similarity between the user feature vector W and the content feature vector to be distributed is calculated for each content to be distributed. Then, as shown in FIG. 18, the contents to be distributed are rearranged according to the predicted value P calculated here.

On the other hand, in S103, when it is determined that it is not a situation reflecting mode, the evaluation value R and sim _(X c, X _i) history content in descending order of the multiplication value of the sorted ranking process is performed (S111) . Then, as described above, the user feature vector is calculated using the history content rearranged in this way, and recommendation information of the content to be distributed is generated.

Here, the ranking process when the situation reflection mode is not employed will be described with reference to FIG. Here, the evaluation value R associated with the history content is multiplied by the similarity sim (X _c , X _i ) between the content feature vector X _c of the designated content being browsed.

  Then, as shown in FIG. 15, the history content rearrangement process is performed according to the multiplication value. Here, the content ID “I0006” having a higher multiplication value is rearranged so as to be higher.

  Next, operational effects of the information distribution server 200 of the present embodiment will be described. In the information distribution server 200 of the present embodiment, the history collection unit 28 acquires content browsed by the user, and the user history management DB 40 acquires browsing history information indicating the history content that is the acquired content for each user. Remember memorize.

Then, when the recommendation request receiving unit 30 receives a recommendation request including a content ID indicating the content currently being browsed, the history similarity calculation unit 21 is identified by the content ID included in the recommendation request ( The similarity sim (X _c , X _i ) between the content feature vector X _c ) and each history content (content feature vector X _i ) stored in the user history management DB 40 is calculated.

  The history evaluation value calculation unit 22 calculates the corrected evaluation value EV value of the history content by multiplying the calculated history similarity by the user evaluation value for the history content. Then, the user feature vector calculation unit 24 rearranges the history in descending order of the corrected evaluation value EV value calculated by the history evaluation value calculation unit 22, and calculates a user feature vector so as to satisfy the arrangement order.

  Then, the content category evaluation value calculation unit calculates a predicted value P indicating the similarity between the user feature vector of the user calculated by the history evaluation value calculation unit 22 and the feature vector of the recommendation target content to be recommended, and the content The transmission unit 32 rearranges the content according to the predicted value P, and determines recommendation information to be recommended content for the user.

  This makes it possible to determine recommendation information that is content suitable for the user based on the content being browsed and history content, and to provide more appropriate content to the user, improving the accuracy of the recommendation information. be able to.

In the information distribution server 200, the recommendation request receiving unit 30 receives a recommendation request including the user's situation vector together with the content ID, and the situation similarity calculating unit 20 receives the user situation vector CS indicating the received current situation. status and status similarity sim indicating the proximity to the situation shown by the previous user situation vector S _n (CS, S _n) calculated, where calculated circumstances similarity sim (CS indicated by, S _n ) Is multiplied by the user's evaluation value R for the history content, and the above-described corrected evaluation value EV obtained by multiplying the history similarity calculated earlier by the user's evaluation value R for the history content. By adding to the value, a corrected evaluation value EV considering the user situation is calculated.

That is, the product of the situation similarity sim (CS, S _n ) and the evaluation value R (may be further multiplied by a predetermined weighting coefficient α), the history similarity sim (X _c , X _i ) and the evaluation value The sum of a multiplication value of R (which may be further multiplied by a predetermined weighting coefficient β) is calculated and used as a corrected evaluation value EV value.

  Then, the user feature vector calculation unit 24 generates a user feature vector according to the corrected evaluation value EV value, and the content category evaluation value calculation unit 26 and the content transmission unit 32 generate recommendation information based on the user feature vector. And can be delivered. Therefore, it is possible to generate appropriate recommendation information according to the current user situation.

DESCRIPTION OF SYMBOLS 10 ... History transmission part, 12 ... Recommendation request transmission part, 14 ... Situation estimation part, 16 ... Recommendation information display part, 20 ... Situation similarity calculation part, 21 ... History similarity calculation part, 22 ... History evaluation value calculation part, 24 ... user feature vector calculation unit, 26 ... content category evaluation value calculation unit, 28 ... history collection unit, 28 ... history collection unit, 28 ... history collection unit, 30 ... recommendation request reception unit, 32 ... content transmission unit, 46 ... Content table, 100 ... Mobile device, 200 ... Information distribution server, 40 ... User history management DB, 42 ... Content management DB, 44 ... User feature vector management DB, 46 ... Content table.

Claims (3)

Browsing content acquisition means for acquiring content specified by the user;
Browsing history information storage means for storing browsing history information indicating history content that is the content acquired by the browsing content acquisition means, for each user;
A recommendation request receiving means for receiving a recommendation request including a content ID indicating the content currently being browsed;
History similarity calculation for calculating the similarity between the content specified by the content ID included in the recommendation request received by the recommendation request receiving means and each history content stored in the browsing history information storage means Means,
History evaluation value calculation means for calculating a history content correction value by multiplying the history similarity calculated by the history similarity calculation means by the user's evaluation value for the history content;
User feature vector calculation means for rearranging the history in descending order of correction evaluation values calculated by the history evaluation value calculation means, and calculating a user feature vector so as to satisfy the arrangement order;
Recommended content for determining recommendation information to be recommended content for a user by calculating the similarity between the user feature vector of the user calculated by the user feature vector calculating means and the feature vector of the recommended content to be recommended A determination means;
A recommendation information generating apparatus comprising: The recommendation request receiving means receives a recommendation request including a user status vector together with a content ID,
Situation vector storage means for storing a user situation vector indicating a past user situation;
The situation similarity indicating the closeness between the situation indicated by the user situation vector indicating the current situation received by the recommendation request receiving means and the situation indicated by the past user situation vector stored in the situation vector storage means And a situation similarity calculation means for calculating
The history evaluation value calculation means further includes a history similarity calculated by the history similarity calculation means by a multiplication value obtained by multiplying the situation similarity calculated by the situation similarity calculation means with a user evaluation value for history content. The recommendation evaluation according to claim 1, wherein a correction evaluation value in consideration of a user situation is calculated by adding the degree to the correction evaluation value obtained by multiplying a degree and a user evaluation value for history content. Information generator. A browsing content acquisition step of acquiring content specified by the user;
A storage step of storing browsing history information indicating history content, which is the content acquired by the browsing content acquisition step, for each user in a browsing history information storage unit;
A recommendation request receiving step for receiving a recommendation request including a content ID indicating the content currently being browsed;
History similarity calculation for calculating the similarity between the content specified by the content ID included in the recommendation request received by the recommendation request receiving step and each history content stored in the browsing history information storage means Steps,
A history evaluation value calculating step of calculating a corrected evaluation value of the history content by multiplying the history similarity calculated by the history similarity calculation step by a user evaluation value for the history content;
A user feature vector calculation step for rearranging the history in the descending order of the correction evaluation value calculated by the history evaluation value calculation step, and calculating a user feature vector so as to satisfy the arrangement order;
Recommended content for determining recommendation information to be recommended content for the user by calculating the similarity between the user feature vector of the user calculated in the user feature vector calculating step and the feature vector of the recommended target content to be recommended A decision step;
A recommendation information generation method comprising:

Images