JP6960838B2 - Information providing equipment, information providing method, and program - Google Patents

Description

The present invention relates to an information providing device, an information providing method, and a program.

Articles are distributed via the network. In relation to this, there is known a technique of recommending related articles that do not substantially overlap in content and that match the tastes and interests of each user as articles related to articles that the user is interested in (for example,). See Patent Document 1).

Japanese Unexamined Patent Publication No. 2010-224623

However, in the conventional technology, the content that is of high interest to the user is recommended based on the behavior history of what kind of content the user has browsed so far, and the user whose behavior history is not sufficiently accumulated. In some cases, it was not possible to recommend content that was of great interest to us.

The present invention has been made in view of the above problems, and an object of the present invention is to provide an information providing device, an information providing method, and a program capable of recommending contents that are presumed to be of interest to the user. There is.

One aspect of the present invention refers to the browsing history of the content distributed via the network, and when the number of times the content is viewed by the target user in a predetermined period is less than or equal to the predetermined number, at least the characteristics of the user other than the target user. Based on this, an acquisition unit that acquires an alternative feature derived as a substitute feature of the target user feature, at least one of the target user feature or the target user alternative feature, and a plurality of contents to be distributed, respectively. An information providing device including a determination unit for determining content recommended to the target user from a plurality of contents to be distributed based on the characteristics of the above.

According to one aspect of the present invention, it is possible to recommend content that is presumed to be of interest to the user.

It is a figure which shows an example of the information providing system 1 including the information providing apparatus 100 in 1st Embodiment. It is a figure which shows an example of the structure of the information providing apparatus 100 in 1st Embodiment. It is a figure which shows an example of the content vector information 134. It is a figure which shows an example of the user information 136. It is a figure which shows an example of the action log 138. It is a figure which shows an example of the user vector information 140. It is a flowchart which shows the flow of a series of processing by an information providing apparatus 100. It is a figure which shows an example of the web page which carries the recommended content. It is a figure which shows other example of the web page which publishes recommended content. It is a figure which shows typically the vector space in which a plurality of other user vectors exist. It is a figure which shows an example of the hardware composition of the information providing apparatus 100 of an embodiment.

Hereinafter, an information providing device, an information providing method, and a program to which the present invention is applied will be described with reference to the drawings.

[Overview]
The information providing device is realized by one or more processors. As a basic function, the information providing device refers to the browsing history of the content distributed via the network, and when the number of times the content is viewed by the target user in a predetermined period exceeds the predetermined number, the content viewed by the target user The characteristics of the target user are derived based on the characteristics, and the characteristics of the derived target user and the characteristics of each of the plurality of contents to be distributed are recommended to the target user from among the plurality of contents to be distributed. It has a function to determine the content.

Further, when the number of times the content is viewed by the target user in a predetermined period is less than or equal to a predetermined number, the information providing device derives an alternative feature that substitutes for the feature of the target user, at least based on the characteristics of the user other than the target user. , It has a function of determining the content recommended to the target user from the plurality of contents to be distributed based on the derived alternative features and the characteristics of each of the plurality of contents to be distributed. As a result, for example, on the premise that there are no user characteristics (for example, zero vector), content that is presumed to be of interest to the user is recommended rather than simply providing content ranked by the number of accesses or the like. can do.

The content in this embodiment is, for example, an article posted on a blog, a website, or the like, and is content including text. Further, the content may be, for example, text, a still image, a moving image, an audio, or the like posted as an advertisement on a medium such as a website.

<First Embodiment>
[overall structure]
FIG. 1 is a diagram showing an example of an information providing system 1 including an information providing device 100 according to the first embodiment. The information providing system 1 in the first embodiment includes, for example, one or more terminal devices 10 and an information providing device 100. These devices are connected via the network NW.

Each device shown in FIG. 1 transmits and receives various information via the network NW. The network NW includes, for example, the Internet, a WAN (Wide Area Network), a LAN (Local Area Network), a provider terminal, a wireless communication network, a wireless base station, a dedicated line, and the like. It should be noted that not all combinations of the devices shown in FIG. 1 need to be able to communicate with each other, and the network NW may partially include a local network.

The terminal device 10 is a terminal device including an input device, a display device, a communication device, a storage device, and an arithmetic device, such as a mobile phone such as a smartphone, a tablet terminal, and various personal computers. The communication device includes a network card such as a NIC (Network Interface Card), a wireless communication module, and the like. In the terminal device 10, a UA (User Agent) such as a web browser or an application program is activated, and a request according to the content input by the user is transmitted to the information providing device 100. Further, the terminal device 10 in which the UA is activated causes the display device to display various images based on the information acquired from the information providing device 100.

The information providing device 100 is, for example, a web server that provides a web page to the terminal device 10 in response to a request from a web browser. The above-mentioned contents are posted on the web page. Further, the information providing device 100 may be an application server that provides contents such as images and sounds to the terminal device 10 in response to a request from the application program. For example, the information providing device 100 is a user who uses a terminal device 10 via a website or an application to perform a service for searching a desired content from a certain database (hereinafter referred to as a search service) such as a document search or an image search. To provide.

[Configuration of information providing device]
FIG. 2 is a diagram showing an example of the configuration of the information providing device 100 according to the first embodiment. As shown in the figure, the information providing device 100 includes, for example, a communication unit 102, a control unit 110, and a storage unit 130.

The communication unit 102 includes, for example, a communication interface such as a NIC. The communication unit 102 communicates with the terminal device 10 via the network NW and acquires various information from the terminal device 10. For example, the communication unit 102 acquires login information from the terminal device 10. The login information is information including a user ID (account ID) and a password associated with an account for using the search service.

The control unit 110 includes, for example, a content vector generation unit 112, a user vector generation unit 114, a recommendation content determination unit 116, a content distribution unit 118, and an information update unit 120. The user vector generation unit 114 is an example of the “deriving unit”, and the recommendation content determination unit 116 is an example of the “acquisition unit” and the “determination unit”.

These components are realized, for example, by a processor such as a CPU (Central Processing Unit) executing a program stored in the storage unit 130. In addition, some or all of the components of the control unit 110 are hardware such as LSI (Large Scale Integration), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array), or GPU (Graphics Processing Unit). It may be realized by (circuit unit; circuitry), or it may be realized by the cooperation of software and hardware.

The storage unit 130 is realized by, for example, a storage device such as an HDD (Hard Disc Drive), a flash memory, an EEPROM (Electrically Erasable Programmable Read Only Memory), a ROM (Read Only Memory), and a RAM (Random Access Memory). In addition to various programs such as firmware and application programs, the storage unit 130 stores content information 132, content vector information 134, user information 136, action log 138, user vector information 140, and the like.

The content information 132 is information including various contents. For example, when the content is a news article, the content information 132 may include text information, images, audio, and the like included in the news article, and information such as a news article distributor and author.

FIG. 3 is a diagram showing an example of the content vector information 134. The content vector information 134 is information in which each content included in the content information 132 is vectorized. For example, a content vector in which the content is vectorized is associated with a content ID for identifying each content. Information.

FIG. 4 is a diagram showing an example of user information 136. The user information 136 is, for example, information in which a user attribute, a password at the time of login, and the like are associated with a user ID (account ID) for identifying each user. User attributes include, for example, demographic attributes (demographic attributes) possessed by the user, such as gender, age, area of residence, income, occupation, and educational background. These demographic attributes shall be pre-registered, for example, when the account is created. Further, in the user attribute, for example, the type of the terminal device 10 indicating whether the terminal device 10 used by the user at the time of login is a mobile phone such as a smartphone, a tablet terminal, or a personal computer. Or, information such as the day and time of login may be included as an attribute.

FIG. 5 is a diagram showing an example of the action log 138. The action log 138 is information based on the browsing history of the content by the user. For example, for each user ID, the content ID of the content browsed by the user and the number of times the content is browsed in a predetermined period are associated with each other. Is. If the user has just registered a new account, the content ID may not be associated with the user ID in the action log 138, for example, USER_1 in the figure. ..

FIG. 6 is a diagram showing an example of the user vector information 140. The user vector information 140 is information in which each user indicated by the user information 136 is vectorized. For example, the user ID is associated with a user vector in which the user is vectorized. In the action log 138, when the content ID is not associated with the user ID, that is, the user who has never browsed the content after registering a new account is assigned to the user ID of that user. On the other hand, the user vector may not be associated (for example, USER_1 in the figure).

The content vector generation unit 112 vectorizes each content included in the content information 132 and generates a content vector. For example, when the content is a news article, the content vector generation unit 112 vectorizes the news article based on the learning model of Countuous Bag-of-Words and / or Skip-gram, using the news article as a corpus. For example, the content vector generation unit 112 divides a news article into a plurality of morphemes, and among the plurality of morphemes, a specific part of speech (noun, etc.) is a learning model of both Countuous Bag-of-Words and Skip-gram. By inputting as a word into a neural network (for example, word2vec) to which is applied, the word vector of each word output from the neural network is acquired. Then, the content vector generation unit 112 uses a value obtained by normalizing the sum of a plurality of word vectors corresponding to each word (for example, dividing by the number of words) as a vector of a news article (an example of a content vector) that is a word extraction source. Derived as.

Further, the content vector generation unit 112 may vectorize each news article by using a learning model such as Bag-of-Words alone or Global Vector Prediction (GloVe). Further, when the content is a still image or a moving image, the content vector generation unit 112 may vectorize the content by using, for example, a convolutional neural network.

The content vector generation unit 112 associates the generated content vector with the content ID of the content that is the source of the content vector, and stores this as the content vector information 134 in the storage unit 130. The process of vectorizing the content may be realized by requesting the external device to perform the process instead of the content vector generation unit 112 performing the process by itself.

The user vector generation unit 114 vectorizes each user included in the user information 136 based on the content vector information 134 and the action log 138, and generates a user vector. For example, the user vector generation unit 114 refers to the action log 138, and acquires the content ID of the content viewed by the user from the storage unit 130 for each user ID. The user vector generation unit 114 refers to the content vector information 134, and acquires the content vector corresponding to the acquired content ID from the storage unit 130. Then, the user vector generation unit 114 derives a set of content vectors corresponding to the content IDs of the contents viewed by the user as the user vector of the user. In the case of the example of FIG. 5, the action log 138 indicates that the user whose user ID is "USER_2" has browsed the content whose content IDs are "CONT_6" and "CONT_14". In such a case, the user vector generation unit 114 refers to the content vector information 134, and corresponds to the content vector V (CONT_6) corresponding to the content whose content ID is "CONT_6" and the content whose content ID is "CONT_14". At least the content vector V (CONT_14) is acquired, and a vector set including at least these content vectors is generated as a user vector of a user whose user ID is "USER_2".

The user vector generation unit 114 associates the generated user vector with the user ID of the user who is the source of the user vector, and stores this as user vector information 140 in the storage unit 130. As described above, in the action log 138, the content ID may not be associated with the user ID. In such a case, the user vector generation unit 114 generates a vector in place of the user vector by a method different from the above-mentioned method. The method of generating this vector will be described later using a flowchart.

The recommendation content determination unit 116 refers to the user vector information 140 stored in the storage unit 130, and among the plurality of user vectors generated by the user vector generation unit 114, the user vector in which the target user is vectorized (hereinafter, the user vector). The target user vector) is acquired from the storage unit 130, and one or more content vectors generated by the content vector generation unit 112 are acquired from the storage unit 130.

Then, the recommendation content determination unit 116 recommends viewing to the target user based on the similarity between the user vector and one or more content vectors acquired from the storage unit 130 (hereinafter, referred to as recommended content). To determine. The target user is, for example, a user who has entered a query in a search window or the like when searching for desired information using a search service after logging in. The target user vector is an example of “characteristics of the target user”.

For example, the recommendation content determination unit 116 derives the cosine similarity with the target user vector for each content vector, and sets the content corresponding to one content vector having the largest cosine similarity with the target user vector into the recommended content. decide. Further, the recommendation content determination unit 116 may determine a set of contents corresponding to each of the upper predetermined number of content vectors having a high cosine similarity with the target user vector as the recommendation content, or may use the target user vector and the target user vector. The set of contents corresponding to each of all the content vectors whose cosine similarity is equal to or higher than the threshold value may be determined as the recommended content.

The content distribution unit 118 uses the communication unit 102 to distribute one or more contents determined as recommended contents by the recommendation content determination unit 116 to the terminal device 10 used for inputting a query by the target user. ..

For example, when the recommended content distributed by the content distribution unit 118 is viewed by the target user, the information update unit 120 may newly add the content ID of the recommended content to the user ID of the target user in the action log 138. The action log 138 is updated by replacing the content ID of the content with the old browsing time and incrementing the counted number of browsing times. At this time, the user vector generation unit 114 may newly generate the target user vector based on the updated action log 138 when the target user vector does not exist in the user vector information 140. Further, the user vector generation unit 114 may update the target user vector based on the updated action log 138 when the target user vector already exists in the user vector information 140.

In the action log 138, the information update unit 120 browses any content within a predetermined period when the content ID exists in the user ID of the target user, that is, when the target user vector already exists. If not, in the action log 138 of the target user, all the content IDs associated with the user IDs may be deleted and the browsing history may be initialized. In this case, the user vector generation unit 114 may erase the already generated target user vector, for example.

[Processing flow]
Hereinafter, the flow of a series of processes by the information providing device 100 will be described according to a flowchart. FIG. 7 is a flowchart showing a flow of a series of processes by the information providing device 100. The processing of this flowchart is performed, for example, when a query for searching the content is further acquired from the terminal device 10 for which the login information has been acquired by the communication unit 102.

First, the user vector generation unit 114 refers to the action log 138, and the user ID included in the login information acquired by the communication unit 102, that is, the number of views associated with the user ID of the target user is a predetermined number or less. It is determined whether or not there is (S100). The predetermined number is, for example, 0 times. In this case, the user vector generation unit 114 refers to the action log 138, and either the target user is a user who has not yet browsed the content after registering the account, or has not browsed any content within a predetermined period and browsed the browsing history. Determines if is an initialized user.

When the user vector generation unit 114 determines that the number of times the target user has viewed the content is not less than or equal to a predetermined number, that is, when the target user is a user who has already viewed the content after registering the account, the recommendation content determination unit 116 With reference to the user vector information 140, the target user vector already generated by the user vector generation unit 114 is acquired, and the cosine similarity between the target user vector and each of the plurality of content vectors is derived (S102).

Next, the recommendation content determination unit 116 is generated by the user vector generation unit 114, among a plurality of user vectors generated based on the content vector corresponding to the content viewed by the user, by the target user vector and the content vector generation unit 112. Based on the similarity with each of the generated one or more content vectors, the recommended content recommended to be viewed by the target user is determined from a set of a plurality of contents corresponding to each content vector (S104).

Next, the content distribution unit 118 uses the communication unit 102 to distribute one or more contents determined as the recommended content by the recommendation content determination unit 116 to the terminal device 10 of the target user (S106).

On the other hand, when the user vector generation unit 114 determines that the number of times the target user has viewed the content is less than or equal to a predetermined number, that is, the target user has not yet viewed the content after registering the account, or within a predetermined period. If the user is a user whose browsing history has been initialized without browsing any of the contents, since the target user vector corresponding to the target user does not exist, an alternative user vector to replace this target user vector is generated (S108). ). The alternative user vector is an example of "alternative features of the target user".

For example, the user vector generation unit 114 refers to the user vector information 140 and derives an average vector obtained by averaging the user vectors of users other than the target user (hereinafter, referred to as other user vectors) as an alternative user vector. For example, it is assumed that there are five user IDs "USER_1" to "USER_1", of which "USER_1" is the user ID of the target user. At this time, if the target user vector V (USER_1) corresponding to the user ID of the target user "USER_1" does not exist, and other user vectors V (USER_1) to V (USER_1) of other users exist, the user. The vector generation unit 114 may generate an alternative user vector using all other user vectors based on the mathematical expression (1). _{V X} in the formula (1) represents an alternative user vector.

Then, the user vector generation unit 114 replaces the generated alternative user vector V _X with the target user vector V (USER_1), so that the target user who has never browsed the content or whose browsing history is initialized is the target. Generate a user vector. In addition, instead of generating the average vector of a plurality of other user vectors as an alternative user vector, the user vector generation unit 114, for example, the similarity of a specific other user vector (for example, the similarity of the other other user vector is more than the threshold value). The average vector of a plurality of other user vectors excluding (also low) may be generated as an alternative user vector, the weighted average vector of a plurality of other user vectors may be generated as an alternative user vector, or a plurality of. A vector that is the median value of another user vector may be generated as an alternative user vector, or an alternative user vector may be generated based on other statistics.

The user vector generation unit 114, instead of generating the substitute user vector V _X with all other users vector, may generate an alternate user vector V _X by using a part of the other users vector ..

For example, the user vector generation unit 114 refers to the user information 136, identifies another user having the same attributes as at least a part of the one or more attributes of the target user, and includes the user vector information 140. Select another user vector of the specified user from all other user vectors. Then, the user vector generation unit 114 generates an alternative user vector using the selected other user vector.

For example, when the user whose user ID is "USER_1" is the target user and it is decided to specify another user who has the same attribute of "gender" as the target user, in the user information 136 illustrated in FIG. Users with user IDs "USER_4" and "USER_5" are identified. In such a case, the user vector generation unit 114 derives a vector obtained by averaging the other user vectors V (USER_4) and V (USER_5) as an alternative user vector based on the mathematical formula (2).

Further, the user vector generation unit 114 may determine the weight when averaging a plurality of other user vectors based on the similarity of the user attributes possessed by other users with respect to the user attributes possessed by the target user.

For example, in the user information 136 illustrated in FIG. 4, when the user whose user ID is "USER_1" is the target user and the users whose remaining user IDs are "USER_2" to "USER_1" are other users. The user vector generation unit 114 derives as an alternative user vector based on the mathematical formula (3).

_{Each of w 2 to w 5} in the mathematical formula (3) represents the weight of each other user vector. In the user information 136 illustrated in FIG. 4, the user whose user ID is "USER_2" does not match the target user at least in terms of "gender", "age", and "region". Further, the user whose user ID is "USER_3" matches the target user in terms of "age" and "region", and does not match the target user in terms of "gender". "Age" is the same when the last digit of the age is ignored and the age is divided by 10 years, for example, teens, 20s, 30s, and so on. For example, since the age of the target user is "45 years old" and the age of the user whose user ID is "USER_3" is "40 years old", the attributes of the "age" of these users match. Further, the user whose user ID is "USER_4" matches the target user for "gender" and does not match the target user for "age" and "region". Further, the user whose user ID is "USER_5" matches the target user in terms of "gender" and "age", and does not match the target user in terms of "region".

_{In such a case, the user vector generation unit 114 has the weight w 3} of the other user vector V (USER_3) of the user whose two attributes match the target user among the three user attributes, and the other user vector V (USER_5). The weight w _{5 of is the largest, and the weight w 4} of the other user vector V (USER_4) of the user whose one attribute matches the target user is made smaller than the weight w ₃ or the weight w ₅ , and both of the target user and the target user. _{The weight w 2} of the other user vector V (USER_2) of the user whose attributes do not match may be minimized.

Next, the recommendation content determination unit 116 derives the cosine similarity between the alternative user vector generated as the target user vector by the user vector generation unit 114 and each of the plurality of content vectors (S110). Then, the recommendation content determination unit 116 proceeds to the process of S104, and determines the content corresponding to each of the one or more content vectors having a high degree of similarity with the alternative user vector as the recommendation content. This ends the processing of this flowchart. In the description of the flowchart described above, it has been described that the alternative user vector generation process (S108) and the recommendation content determination process (S104) are performed in the same series of processing flows, but the present invention is limited to this. I can't. For example, the process of generating the alternative user vector and the process of determining the recommended content may be performed independently of each other. In this case, it is assumed that the alternative user vector is generated in advance when the recommendation content is determined. That is, the recommendation content determination unit 116 shall acquire the alternative user vector already generated from the storage unit 130 in the process of S108.

FIG. 8 is a diagram showing an example of a web page on which recommended content is posted. For example, the content distribution unit 118 causes one or more recommended contents to be posted in a part area (Ra in the figure) of the web page corresponding to the query input by the target user, and distributes this web page to the terminal device 10 of the target user. By providing, the recommended content may be distributed.

According to the first embodiment described above, when the number of views of the content included in the action log 138 of the target user exceeds a predetermined number in the browsing history of one or more contents distributed via the network, the target is targeted. Based on the content vector showing the characteristics of one or more contents viewed by the user, the target user vector showing the characteristics of the target user is generated, and the content is browsed for a predetermined period included in the browsing history of the content of the target user. When the number of times is less than or equal to a predetermined number, a user vector generation unit 114 that generates an alternative user vector as a substitute for the target user vector based on another user vector indicating the characteristics of one or more users other than the target user, and a user vector. Based on at least one of the target user vector or the alternative user vector generated by the generation unit 114 and the content vector corresponding to each of the one or more contents, the target user is recommended from the set of one or more contents. By providing the recommendation content determination unit 116 that determines the recommendation content, it is possible to recommend the content that is presumed to be of interest to the user.

For example, when the recommended content is not determined by using the user vector of a user other than the target user as in the present embodiment, it is generally called the most popular page view that does not depend on the user's action history, and the most page view at that time. Popular contents with a large number are delivered preferentially, and contents are delivered in the order of new arrival. However, new content is not ranked high in distribution priority because it has a poor track record of being viewed by users, and is not distributed as recommended content. On the other hand, when the recommended content such as a news article or an advertisement is determined by using the average vector of the user vector of a user other than the target user as in the present embodiment, as a result, popular content is delivered. However, even content that has not been viewed by users, such as new content, can be distributed as recommended content.

<Second Embodiment>
Hereinafter, the second embodiment will be described. The second embodiment is different from the above-described first embodiment in that the user vector generation unit 114 generates a substitute user vector even when the number of views of the target user exceeds a predetermined number. Hereinafter, the differences from the first embodiment will be mainly described, and the points common to the first embodiment will be omitted. In the description of the second embodiment, the same parts as those of the first embodiment will be described with the same reference numerals.

When the number of views of the target user exceeds a predetermined number, the user vector generation unit 114 in the second embodiment generates an alternative user vector based on the other user vector of the other user, as in the first embodiment described above. Generate. Then, the user vector generation unit 114 generates a new alternative user vector as a substitute for the target user vector based on the alternative user vector based on the other user vector and the target user vector that has already been generated for the target user. do. The already generated target user vector is, for example, a set vector of content vectors corresponding to each of one or more contents viewed by the target user.

For example, the user vector generation unit 114 generates a new alternative user vector Vx # as a substitute for the target user vector of a certain target user k based on the mathematical formula (4). The new alternative user vector Vx # is another example of an "alternative feature".

_{Λ of the formula (4) represents the weight of the alternative user vector V X} generated based on the other user vector of other users excluding the target user k, and (1-λ) is the already generated target user k. Represents the weight of the target user vector V (USER_k) of. For example, the weight λ increases as the number of views of the content of the target user k increases, and decreases as the number of views decreases. That is, when the user vector generation unit 114 _{derives the weighted average of the alternative user vector V X} and the target user vector V (USER_k), as the number of times the content of the target user k is viewed increases, the new alternative user vector The contribution of the existing target user vector V (USER_k) to Vx # is increased, and the contribution _{of the alternative user vector V X to} the new alternative user vector Vx # is increased as the number of times the content of the target user k is viewed decreases. ..

For example, in the action log 138 illustrated in FIG. 5, when the user whose user ID is "USER_1" is the target user, this target user is viewed significantly more times than the user other than the user whose user ID is "USER_1". few. In such a case, since the number of contents viewed by the target user is small, it is assumed that the target user vector V (USER_3) of the target user does not have sufficient reliability as a vector representing the characteristics and properties of the target user. Will be done. Therefore, when the number of views is small and the target user vector has low reliability as a vector representing the characteristics and properties of the target user, the user vector generation unit 114 generates an alternative user vector as an average vector of other user vectors. The contribution of V _X is increased, and the contribution of the existing target user vector V (USER_k) is increased while decreasing the contribution of the _{alternative user vector V X} as the number of views increases.

The recommendation content determination unit 116 in the second embodiment includes the average vector of the past alternative user vector Vx and the existing target user vector V (USER_k) generated as a new alternative user vector Vx # by the user vector generation unit 114. , The similarity with each of the plurality of content vectors is derived, and one or more contents corresponding to the content vectors having a large similarity with the average vector are determined as the recommended contents.

According to the second embodiment described above, the target is a set sum of content vectors in which the weight is increased as the number of times the target user has viewed the content in the past is smaller, and the weight is decreased as the number of times the target user has viewed the content is larger. An average vector (that is, a new alternative user vector Vx #) of the user vector and the alternative user vector which is the average vector of the other user vector generated as a substitute for the target user vector is generated, and this average vector and each content vector are used. In order to obtain the similarity of the content and determine one or more contents corresponding to the content vector having a large similarity with the average vector as the recommended content, for example, a sufficient time has passed since the target user registered the account. Even if the number of content vectors used to generate the target user vector is not sufficient, the target user will be interested in considering the tendency of the content that many other users have viewed in the past. Content that is presumed to be can be recommended.

<Third Embodiment>
Hereinafter, the third embodiment will be described. In the third embodiment, the target user vector, which is the union of a plurality of content vectors corresponding to each of the one or more contents viewed by the target user, and the average vector of other user vectors generated in place of the target user vector. It differs from the first and second embodiments described above in that the similarity with the content vector is obtained for each of the alternative user vectors, and the recommended content is determined for them. Hereinafter, the differences from the first and second embodiments will be mainly described, and the points common to the first and second embodiments will be omitted. In the description of the third embodiment, the same parts as those of the first and second embodiments will be described with the same reference numerals.

The user vector generation unit 114 in the third embodiment generates a set vector of a plurality of content vectors corresponding to each of the one or more contents viewed by the target user as the target user vector, and substitutes for the target user vector. As an alternative user vector, the average vector of the other user vector is generated.

The recommendation content determination unit 116 in the third embodiment is a total of 2 of the target user vector, which is a set vector of a plurality of content vectors generated by the user vector generation unit 114, and the alternative user vector, which is an average vector of other user vectors. For each of the two vectors, the similarity with the content vector is derived. That is, the recommendation content determination unit 116 uses the alternative user vector, which is the average of the user vectors of many other users, even when the content has been viewed a sufficient number of times and a highly reliable user vector is generated. Derivation of similarity with the content vector.

Then, the recommendation content determination unit 116 determines the content corresponding to each of the one or more content vectors having a high degree of similarity with the target user vector as the first type of recommended content, and the degree of similarity with the alternative user vector is high. The content corresponding to each of one or more large content vectors is determined as the second type of recommended content.

The content distribution unit 118 in the third embodiment distributes two types of recommended contents determined by the recommendation content determination unit 116 to the terminal device 10 of the target user by using the communication unit 102.

FIG. 9 is a diagram showing another example of a web page on which recommended content is posted. For example, the content distribution unit 118 posts one type of recommended content in each of two different areas (Rb and Rc in the figure) of the web page area corresponding to the query input by the target user, and this web. The recommended content may be distributed by providing the page to the terminal device 10 of the target user.

For example, in the case of the first type of recommended content corresponding to the target user vector which is a set vector of a plurality of content vectors, the area Rb where the recommended content is posted includes the recommended content and "recommended content for you". Such characters indicating that the recommended content is recommended to the target user may be displayed. Further, in the case of the second type of recommended content corresponding to the alternative user vector which is the average vector of the other user vector, the area Rc where the recommended content is posted includes the recommended content and "content recommended to other users". Characters indicating that the recommended content is recommended to other users, such as, may be displayed. In the above-mentioned example, the first type of recommended content corresponding to the target user vector and the second type of recommended content corresponding to the alternative user vector are displayed separately, but the present invention is not limited to this. These two types of recommended contents may be displayed mixed with each other without distinction.

According to the third embodiment described above, a total of two vectors, a target user vector which is a set vector of a plurality of content vectors and an alternative user vector which is an average vector of other user vectors, are referred to as content vectors. Content that is presumed to be of interest to the target user because the similarity is derived, the recommended content is determined based on the similarity with the target user vector, and the recommended content is determined based on the similarity with the alternative user vector. And the content that is presumed to be of interest to a large number of users can be delivered to the terminal device 10 of the target user.

<Fourth Embodiment>
Hereinafter, the fourth embodiment will be described. In the fourth embodiment, a plurality of other user vectors are scattered in a vector space stretched by a basis vector (that is, a content vector) that is the basis of these user vectors, and other user vectors that are densely packed in the vector space. It differs from the first to third embodiments described above in that an alternative user vector is generated using only. Hereinafter, the differences from the first to third embodiments will be mainly described, and the points common to the first to third embodiments will be omitted. In the description of the fourth embodiment, the same parts as those of the first to third embodiments will be described with the same reference numerals.

FIG. 10 is a diagram schematically showing a vector space in which a plurality of other user vectors exist. In the illustrated example, a two-dimensional vector space is schematically shown for the sake of brevity. For example, the user vector generating unit 114 in the fourth embodiment obtains the distance between the plurality of other users vector interspersed in a vector space S _a, clustering the other users between vectors that are within a certain distance. The distance of the other user vector may be represented by, for example, the cosine similarity. The user vector generation unit 114 extracts from the vector space S _{a a subspace S b in} which a cluster having the largest number of other user vectors exists among one or more clusters in which a set of a plurality of other user vectors is clustered. Then, the user vector generation unit 114 derives the average vector of a plurality of other user vectors included _{in the extracted subspace S b} , that is, the center of gravity vector of the _{subspace S b as an alternative user vector.}

According to the fourth embodiment described above, since the alternative user vector is generated using only the other user vectors that are densely packed in the vector space, the interest or interest is relatively among a plurality of users other than the target user. After excluding the user vector of the user who is far from other users, the similarity between the alternative user vector and the content vector can be obtained. As a result, the content that is presumed to be of interest to the target user can be determined more accurately.

<Hardware configuration>
The information providing device 100 of the above-described embodiment is realized by, for example, a hardware configuration as shown in FIG. FIG. 11 is a diagram showing an example of the hardware configuration of the information providing device 100 of the embodiment.

The information providing device 100 includes NIC100-1, CPU100-2, RAM100-3, ROM100-4, a secondary storage device 100-5 such as a flash memory or an HDD, and a drive device 100-6 having an internal bus or a dedicated communication line. The configuration is interconnected by. A portable storage medium such as an optical disk is mounted on the drive device 100-6. A program stored in a portable storage medium mounted on the secondary storage device 100-5 or the drive device 100-6 is expanded into the RAM 100-3 by a DMA controller (not shown) or the like, and executed by the CPU 100-2. As a result, the control unit 110 is realized. The program referred to by the control unit 110 may be downloaded from another device via the network NW.

Although the embodiments for carrying out the present invention have been described above using the embodiments, the present invention is not limited to these embodiments, and various modifications and substitutions are made without departing from the gist of the present invention. Can be added.

1 ... Information providing system, 10 ... Terminal device, 100 ... Information providing device, 102 ... Communication unit, 110 ... Control unit, 112 ... Content vector generation unit, 114 ... User vector generation unit, 116 ... Recommended content determination unit, 118 ... Content distribution department, 120 ... Information update department, 130 ... Storage department

Claims (15)

Referring to browsing history content distributed through a network, if Views of contents in a predetermined period of the target user is below a predetermined number, and based on the characteristics of at least the target user other than the user, the target user A derivation unit that derives alternative features that replace the features of
Content recommended to the target user from among the plurality of contents to be distributed based on at least one of the characteristics of the target user or an alternative feature of the target user and each of the characteristics of the plurality of contents to be distributed. With a decision department to decide ,
The derivation unit further refers to the browsing history of the content, and when the number of times the content is viewed by the target user in a predetermined period exceeds a predetermined number, the derivation unit further refers to the content browsed by the target user based on the characteristics of the content. Derivation of the characteristics of the target user,
Information providing apparatus. With reference to the browsing history of the content distributed via the network, when the number of times the content is viewed by the target user in a predetermined period is less than or equal to the predetermined number, at least based on the characteristics of the user other than the target user, the target user A derivation unit that derives alternative features that substitute for features,
Content recommended to the target user from among the plurality of contents to be distributed based on at least one of the characteristics of the target user or an alternative feature of the target user and each of the characteristics of the plurality of contents to be distributed. With a decision department to decide,
The derivation unit derives the weighted average of the feature of the target user and the substitute feature of the target user as the substitute feature of the target user.
Information providing device . The derivation unit derives a statistic for a feature of a user other than the target user as an alternative feature of the target user.
The information providing device according to claim 1 or 2. The derivation unit
From the characteristics of users other than the target user, the characteristics of the user to which the attribute matching the attribute of the target user is given are selected.
Derivation of the alternative feature based on the feature of the selected user.
The information providing device according to any one of claims 1 to 3. The derivation unit
Based on the similarity of the attributes of users other than the target user to the attributes of the target user, the characteristics of the users other than the target user are weighted.
Derivation of alternative features of the target user based on the weighted features of users other than the target user.
The information providing device according to any one of claims 1 to 4. The derivation unit increases the weight as the similarity increases.
The information providing device according to claim 5. The derivation unit further derives an alternative feature of the target user based on the characteristics of a user other than the target user and the characteristics of the target user.
The information providing device according to any one of claims 1 to 6. The determination unit is based on the characteristics of a user other than the target user, the alternative characteristics derived based on the characteristics of the target user by the derivation unit, and the respective characteristics of the plurality of contents to be distributed. Determine the content recommended to the target user from a plurality of contents to be distributed.
The information providing device according to claim 7. The derivation unit reduces the weight of the characteristics of the target user as the number of times the content is viewed in a predetermined period of the target user is smaller.
The information providing device according to claim 2. The determination unit further determines the content recommended to the target user from among the plurality of contents to be distributed, based on the characteristics of the target user and the characteristics of the plurality of contents to be distributed. do,
The information providing device according to any one of claims 1 to 9. A distribution unit that distributes the content determined by the determination unit to the terminal device of the target user is further provided.
The information providing device according to any one of claims 1 to 10. The computer
Referring to browsing history content distributed through a network, if Views of contents in a predetermined period of the target user is below a predetermined number, and based on the characteristics of at least the target user other than the user, the target user Derivation of alternative features to replace the features of
Content recommended to the target user from among the plurality of contents to be distributed based on at least one of the characteristics of the target user or an alternative feature of the target user and each of the characteristics of the plurality of contents to be distributed. to determine,
With reference to the browsing history of the content, when the number of times the content is viewed by the target user in a predetermined period exceeds a predetermined number, the characteristics of the target user are derived based on the characteristics of the content viewed by the target user. ,
Information provision method. The computer
With reference to the browsing history of the content distributed via the network, when the number of times the content is viewed by the target user in a predetermined period is less than or equal to the predetermined number, at least based on the characteristics of the user other than the target user, the target user Derivation of alternative features to replace the features,
Content recommended to the target user from among the plurality of contents to be distributed based on at least one of the characteristics of the target user or an alternative feature of the target user and each of the characteristics of the plurality of contents to be distributed. Decide,
A weighted average of the target user's feature and the target user's alternative feature is derived as the target user's alternative feature.
Information provision method. On the computer
Referring to browsing history content distributed through a network, if Views of contents in a predetermined period of the target user is below a predetermined number, and based on the characteristics of at least the target user other than the user, the target user To derive an alternative feature that replaces the feature of
Content recommended to the target user from among the plurality of contents to be distributed based on at least one of the characteristics of the target user or an alternative feature of the target user and each of the characteristics of the plurality of contents to be distributed. Let me decide
With reference to the browsing history of the content, when the number of times the content is viewed by the target user in a predetermined period exceeds a predetermined number, the characteristics of the target user are derived based on the characteristics of the content viewed by the target user. ,
program. On the computer
With reference to the browsing history of the content distributed via the network, when the number of times the content is viewed by the target user in a predetermined period is less than or equal to the predetermined number, at least based on the characteristics of the user other than the target user, the target user Derivation of alternative features to substitute for features,
Content recommended to the target user from among the plurality of contents to be distributed based on at least one of the characteristics of the target user or an alternative feature of the target user and each of the characteristics of the plurality of contents to be distributed. Let me decide
A weighted average of the target user's feature and the target user's alternative feature is derived as the target user's alternative feature.
program.

Images