Information processing device

Abstract

This server 10 comprises: a display control unit 14 for causing a link image IM to be displayed in a top screen SC1; a first model generation unit 161 for generating, by machine learning based on information that indicates whether or not the link image IM is selected by a user, the user attribute of the user, and information that indicates the content displayed as the link image IM, a first model M1 which is constituted so as to output for each content a first score that pertains to the probability of the link image IM being selected by a user; a user attribute acquisition unit 12 for acquiring the user attribute of a relevant user; and an image determination unit 131 for acquiring the first score for each content by inputting the user attribute of the relevant user to the first model M1 and determining the image of a content the first score of which is high as the image associated with the link image IM in the top screen SC1, that is preferentially presented to the relevant user.

Description

[Technical field]

[0001] One aspect of the present invention relates to an information processing device. [Background technology]

[0002] In a content providing service that provides (sells) content such as videos and electronic books to users, a mechanism is known that presents a user with a screen displaying a list of content recommended to the user. For example, icons corresponding to each content are displayed aligned vertically and horizontally on one screen. When there are many contents to be recommended, it may not be possible to display icons corresponding to all the contents on one screen. In such a case, a link image or the like may be used to transition to another screen for displaying a list of contents that do not fit on one screen (contents that were not displayed on the first screen). By selecting the link image displayed on the first screen, the user can open the other screen and access the list of contents that were not displayed on the first screen. For example, an image or the like showing the contents of the content may be associated with such a link image (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2018-180612 A Summary of the Invention [Problem to be solved by the invention]

[0004] In the content provision services described above, it is expected that the conversion rate (the rate at which users use or purchase content) will be improved by exposing more content to users. For this reason, in the system using link images as described above, a system is required that increases the probability that users will select a link image (hereinafter referred to as "selection probability") in order to expose more content to users.

[0005] Therefore, an object of one aspect of the present invention is to provide an information processing device capable of improving the conversion rate by increasing the selection probability of a link image. [Means for solving the problem]

[0006] An information processing device according to one aspect of the present invention includes a display control unit that displays, in a first screen presented to a user, a list of some of a plurality of contents, and a link image in which a link to a second screen in which a list of hidden contents not displayed in the first screen is displayed and an image of one or more hidden contents is associated; a first model generation unit that generates a first model by machine learning based on information indicating whether the link image has been selected by the user, the user attributes of the user, and information indicating the content displayed as the link image, the first model being configured to output, for each piece of content, a first score corresponding to user attributes relating to the probability that the link image will be selected by the user when an image of the content is associated with the link image; a user attribute acquisition unit that acquires user attributes of a target user to whom the first screen is presented; and an image determination unit that acquires a first score for each piece of content by inputting the user attributes of the target user acquired by the user attribute acquisition unit into the first model, and preferentially determines images of content with high first scores as images to be associated with the link images in the first screen presented to the target user.

[0007] In an information processing device according to one aspect of the present invention, when there are a plurality of contents that cannot be displayed on one screen (first screen), a link image having a link function to a second screen for displaying non-displayed contents that are not displayed on the first screen is displayed on the first screen. Also, a first model is generated that inputs user attributes and outputs a first score related to the selection probability. By preferentially associating images of contents with high first scores obtained by such a first model with the link image, the selection probability of the link image can be improved. As a result, it is possible to expose more contents to the user's eyes, and the conversion rate can be improved. Effect of the Invention

[0008] According to one aspect of the present invention, it is possible to provide an information processing device capable of improving a conversion rate by increasing the selection probability of a link image. [Brief description of the drawings]

[0009] [Figure 1] FIG. 2 is a diagram illustrating an overall configuration of a server according to an embodiment. [Diagram 2] 11A and 11B are diagrams illustrating an example of a top screen and a genre details screen. [Diagram 3] FIG. 11 illustrates an example of a link image generating process. [Figure 4] 10 is a flowchart illustrating an example of an operation of the server. [Diagram 5] FIG. 2 illustrates an example of a hardware configuration of a server. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, an embodiment of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same or corresponding elements are designated by the same reference numerals, and duplicated explanations will be omitted.

[0011] FIG. 1 is a diagram showing an overall configuration of a server 10 according to an embodiment. The server 10 is an information processing device for supporting a recommendation system that provides a conventionally known recommendation service. A recommendation service is a service that presents a list of recommended content to a user in a content providing service that provides (sells) content such as videos and books to users. The server 10 may be a system separate from the recommendation system, or may also serve as the recommendation system.

[0012] As an example, as shown in (A) of FIG. 2, when a user operates the user terminal 20 to access a content providing service, a top screen SC1 (first screen) of the content providing service is displayed on the display unit 21 of the user terminal 20. A list of contents C1 recommended to the user is displayed on the top screen SC1. In this embodiment, the user terminal 20 is a smartphone equipped with a touch panel display as the display unit 21. However, the user terminal 20 is not limited to the above form, and may be any terminal having the display unit 21, for example, a tablet terminal, a desktop PC, a laptop PC, etc.

[0013] In this embodiment, the content C1 to be recommended is displayed for each content genre. In the example of FIG. 2, two contents C1 ("Drama A" and "Drama B") belonging to the genre "drama", two contents C1 ("Anime A" and "Anime B") belonging to the genre "anime", and two contents C1 ("Movie A" and "Movie B") belonging to the genre "movie" are displayed on the top screen SC1. As an example, each content C1 displayed on the top screen SC1 has the form of a rectangular icon associated with an image of each content C1. The image of the content is an image showing the contents of the content (for example, the title screen of a video content, a screen showing the cover of a book content, etc.).

[0014] The user can use (for example, view, purchase, etc.) the content C1 in which he or she is interested by performing an operation of selecting the icon of the content C1 (for example, a touch operation, a click operation, etc.).

[0015] In each genre, there may be contents other than the content C1 displayed on the top screen SC1 that should be recommended to the user. However, since the size of the display area of ​​the display unit 21 of the user terminal 20 is limited, there are cases where all contents to be recommended cannot be displayed on the top screen SC1. That is, in the example of FIG. 2, only the contents C1 ranked first and second in the recommendation degree to the user in each genre are displayed on the top screen SC1, and contents ranked third and lower are not displayed. For this reason, the top screen SC1 displays a link image IM associated with a link to a genre details screen SC2 (second screen) that displays a list of contents C2 (hidden contents) that could not be displayed on the top screen SC1 for each genre. The link image IM is associated with one or more images of the content C2. The recommendation degree as described above can be calculated by a recommendation system using a known recommendation mechanism.

[0016] As shown in FIG. 2B, for example, when a link image IM ("Link 1") of the genre "drama" is selected, the screen transitions from the top screen SC1 to a genre details screen SC2 displaying a list of contents C2 ("Drama C" to "Drama K" in the example of FIG. 2) that are recommended contents belonging to the genre "drama" but were not displayed on the top screen SC1. The same is true when a link image IM ("Link 2", "Link 3") of another genre is selected. Note that in the example of FIG. 2B, the list of contents displayed on the genre details screen SC2 does not include the content C1 that was displayed on the top screen SC1, but the genre details screen SC2 may display the content C2 that was not displayed on the top screen SC1 together with the content C1 that was displayed on the top screen SC1.

[0017] In a content providing service, it is expected that the conversion rate will be improved by exposing more content to the user's eyes. To achieve this, it is effective to improve the selection probability of the link image IM (i.e., the probability of guiding the user to the genre details screen SC2). Therefore, the server 10 is configured to execute a process of personalizing the link image IM according to the attributes of the user in order to improve the selection probability of the link image IM. The configuration of the server 10 will be described in detail below.

[0018] 1, the server 10 includes a request receiving unit 11, a user attribute acquiring unit 12, a link image setting unit 13, a display control unit 14, a user log acquiring unit 15, and a model generating unit 16. The server 10 also includes a user attribute storage unit 10a, a content ID storage unit 10b, and a user log storage unit 10c as elements for storing various data.

[0019] The request receiving unit 11 receives a request for access to the content providing service from a user (user terminal 20). More specifically, the request receiving unit 11 receives from the user terminal 20 a request for information on a top screen SC1 of the content providing service.

[0020] The user attribute acquisition unit 12 acquires user attributes of a target user to whom the top screen SC1 is presented. In this embodiment, the user attribute storage unit 10a stores user attributes for each user in advance. As an example, the user attribute storage unit 10a stores a user attribute associated with a user ID for uniquely identifying a user (in this embodiment, a user attribute vector z described later). The user attribute acquisition unit 12 acquires a user attribute vector z corresponding to the user ID of the target user from the user attribute storage unit 10a. Here, the target user is a user of the user terminal 20 that is the transmission source of the information request accepted by the request acceptance unit 11. For example, the user attribute acquisition unit 12 identifies the user ID of the target user based on information for identifying the user included in the information request (for example, a terminal ID associated with the user ID, etc.).

[0021] User attributes are information related to the characteristics (attributes) of a user. User attributes may include, for example, profile information such as the user's age (or generation), gender, address, etc. User attributes may also include information related to the user's hobbies and preferences estimated from, for example, a questionnaire survey or the above-mentioned profile information. Information related to the user's hobbies and preferences is, for example, a preference for each genre of content (for example, an index that takes a larger value as the user's preference increases).

[0022] The user attribute storage unit 10a stores a user attribute vector z, which is a numerical value and vectorization of the various items described above, for each user ID. The user attribute vector z is generated, for example, as follows. As an example, consider a case where the user attribute vector z is expressed as a one-dimensional vector (age, gender, address, ...) including the items of age, gender, and address. In this case, by appropriately labeling (quantifying) each item based on a predetermined rule (for example, an agreement such as setting male to "0" and female to "1" for gender), a numerical vector such as (0.4, 1.0, 0.7, ...) is obtained as the user attribute vector z of a user having a user attribute such as (40s, female, Kyushu, ...).

[0023] The link image setting unit 13 sets a link image IM personalized for the target user based on the user attribute vector z of the target user acquired by the user attribute acquisition unit 12. The link image setting unit 13 has an image determination unit 131 and a frame interval determination unit 132. The link image setting unit 13 also has a first model M1 used in the processing of the image determination unit 131 and a second model M2 used in the processing of the frame interval determination unit 132.

[0024] The image determination unit 131 determines which image of the content C2 is to be displayed in the link image IM presented to the target user. To this end, the image determination unit 131 first inputs the user attribute vector z of the target user to the first model M1 to obtain a first score for each content C2. For example, the image determination unit 131 refers to the content ID storage unit 10b that stores a content ID for uniquely identifying each content, thereby obtaining the content IDs of all the contents C2 that are not displayed on the top screen SC1, and obtaining the first score corresponding to each content ID.

[0025] The first model M1 is a machine learning model generated (including a case where it is updated; the same applies below) by the first model generation unit 161 described later. More specifically, the first model M1 is a model obtained by performing reinforcement learning in which a reward is given to a piece of content when a link image IM is selected by a user in a situation where an image of the content is displayed as the link image IM. As an example, the first model M1 is obtained using a bandit algorithm (for example, a contextual bandit algorithm such as LinUCB), which is a type of reinforcement learning algorithm. As an example, the first model M1 can be expressed by the following (Equation 1). The first model M1 is obtained by using a parameter (θ a and A a ) The parameter θ a and A a are parameters that are appropriately updated by reinforcement learning. Details of these parameters (initial values ​​and update methods) will be described later.

number

[0026] In (Formula 1), "a" is an identifier indicating the content ID. a " indicates the first score of content a (content corresponding to content ID "a"). "z" indicates the above-mentioned user attribute vector. "θ a” is a unit vector of the same dimension as the user attribute vector z. θ a The higher the probability that a user (a user with user attribute vector z) will select the link image IM when the image of content a is displayed as the link image IM, the higher the value of "θ a T z" is learned (updated) so that "θ a T "z" is a term that contributes to "exploitation" in the bandit algorithm (the exploitation term).

[0027] The second term in (Equation 1) is the term that contributes to "search" in the bandit algorithm (search term). "α" is the weight of the search term and is determined arbitrarily. By adjusting the size of α, the ratio of exploitation to search can be adjusted.

[0028] The image determination unit 131 calculates a first score for each piece of content C2 based on the user attribute vector z and the first model M1 corresponding to each piece of content C2 (that is, (Equation 1) corresponding to each piece of content C2).

[0029] Next, the image determination unit 131 preferentially determines an image of the content C2 having a high first score as an image to be associated with the link image IM in the top screen SC1 presented to the target user. In this embodiment, the image determination unit 131 determines an image to be associated with the link image IM for each genre. Furthermore, the image determination unit 131 determines each image of a plurality (N pieces) of contents C2 as an image to be associated with the link image IM. N is a preset value equal to or greater than 2. More specifically, the image determination unit 131 ranks the plurality of contents C2 in descending order of the first score for each genre, and determines each image of the contents C2 from the first place to the Nth place as an image to be associated with the link image IM.

[0030] In this embodiment, the link image IM is configured as a GIF image (animation image) in which each image of the N pieces of content C2 determined by the image determination unit 131 is switched in sequence at a fixed frame interval. The frame interval determination unit 132 determines the length of the frame interval to be applied to the link image IM presented to the target user. For this purpose, first, the frame interval determination unit 132 inputs the user attribute vector z of the target user to the second model M2 to obtain a second score for each length of the frame interval.

[0031] The second model M2 is a machine learning model generated (including a case where it is updated; the same applies below) by the second model generation unit 162 described later. More specifically, the second model M2 is a model obtained by executing reinforcement learning that gives a reward to a length of a frame interval when a link image IM is selected by a user under a situation in which a certain length of a frame interval is applied to the link image IM. As an example, the second model M2 is obtained using a bandit algorithm (e.g., a contextual bandit algorithm such as LinUCB) which is a type of reinforcement learning algorithm, similar to the first model M1. As an example, the second model M2 can be expressed by the following (Equation 2). The second model M2 is obtained by using a parameter (ρ t and B. t ) The parameter ρ t and B. t are parameters that are appropriately updated by reinforcement learning. Details of these parameters (initial values ​​and update methods) will be described later.

number

[0032] In (Equation 2), "t" is an identifier indicating a candidate for the length of the frame interval. t " denotes the second score of the frame interval length t. "z" denotes the user attribute vector described above. "ρt ” is a unit vector of the same dimension as the user attribute vector z. t When the frame interval length t is applied to the link image IM, the higher the probability that a user (a user with user attribute vector z) selects the link image IM, the higher the probability of the link image IM being selected. t T z" is learned (updated) so that "ρ t T "z" is a term that contributes to "exploitation" in the bandit algorithm (exploitation term).

[0033] The second term in (Equation 2) is the term that contributes to "search" in the bandit algorithm (search term). "β" is the weight of the search term and is determined arbitrarily. By adjusting the size of β, the ratio of exploitation to search can be adjusted.

[0034] The frame interval determination unit 132 calculates a second score for each candidate frame interval length based on the user attribute vector z and a second model M2 corresponding to each candidate frame interval length (i.e., (Equation 2) corresponding to each candidate frame interval length).

[0035] Next, the frame interval determination unit 132 determines, with priority given to the length of the frame interval having a higher second score, as the length of the frame interval to be applied to the link image IM in the top screen SC1 presented to the target user. For example, the frame interval determination unit 132 determines the length of the frame interval having the maximum second score as the length of the frame interval to be applied to the link image IM.

[0036] The display control unit 14 displays a list of some of the multiple contents C1 and a link image IM in a top screen SC1 presented to the target user, as shown in Fig. 2. The link image IM is associated with a link to a genre detail screen SC2 displaying a list of contents C2 not displayed in the top screen SC1, and an image of one or more contents C2 (i.e., the content C2 determined by the image determination unit 131).

[0037] The display control unit 14 displays, on the top screen SC1, a list of some of the contents C1 associated with each of a plurality of content genres (in the example of FIG. 2, drama, anime, and movie) and link images IM for each genre (in the example of FIG. 2, link 1, link 2, and link 3).

[0038] Furthermore, the display control unit 14 changes the image displayed as the link image IM in accordance with the passage of time. More specifically, the display control unit 14 changes the image displayed as the link image IM at a fixed frame interval (i.e., the length of the frame interval determined by the frame interval determination unit 132). For example, the display control unit 14 displays, on the top screen SC1 presented to the target user, a link image IM configured so that images of the contents C2 ranked 1st to Nth in descending order of the first score are switched at a fixed frame interval.

[0039] For example, if the length of the frame interval applied to the link image IM is 1 second, the image of the first-place content C2 is displayed in the link image IM first, then 1 second later the image of the second-place content C2 is displayed, 2 seconds later the image of the third-place content C2 is displayed, and N-1 seconds later the image of the Nth-place content C2 is displayed. After the image of the Nth-place content C2 is displayed as the link image IM, the image of the first-place content C2 is displayed again in the link image IM. In this way, the images of the first-place to Nth-place contents C2 are displayed in a loop in the link image IM.

[0040] The display control unit 14 transmits data for displaying the link image IM to the user terminal 20 of the target user, thereby causing the link image IM as described above to be displayed on the top screen SC1 displayed on the display unit 21 of the user terminal 20.

[0041] The display control unit 14 may generate a GIF image in which the images of the first to Nth ranked contents C2 are switched in this order at regular frame intervals, and transmit data of the GIF image to the user terminal 20 as data for displaying the link image IM. However, since it takes a certain amount of time to generate a GIF image, there is a risk of a time lag occurring between when the top screen SC1 (part other than the link image IM) is displayed on the user terminal 20 and when the link image IM is properly displayed. In order to eliminate such a time lag, the display control unit 14 may perform the following process.

[0042] The display control unit 14 may first transmit to the user terminal 20 an image to be displayed first as a link image IM among the images of the plurality of contents C2 (the first to Nth contents) determined by the image determination unit 131. After that, the display control unit 14 may transmit to the user terminal 20 a GIF image (animation image) configured so that the images of the plurality of contents C2 (the first to Nth contents) change over time. Note that the images of each content are stored in the server 10 (for example, in the content ID storage unit 10b) in association with, for example, a content ID. In this case, the display control unit 14 can acquire the image of each content C2 based on the content ID of each of the first to Nth contents C2 determined by the image determination unit 131.

[0043] The above-mentioned process of the display control unit 14 will be specifically described with reference to FIG. 3. In the example of FIG. 3, the image determination unit 131 determines the content C003 (content with the content ID "C003"; hereinafter, the same notation is used), the content C006, C002, ..., and the content C021 as the first to Nth contents C2. In this case, the display control unit 14 may first transmit to the user terminal 20 only data of an image (hereinafter, referred to as "C003 image") corresponding to the content C003 to be displayed in the link image IM. Then, the display control unit 14 may instruct the user terminal 20 to generate and present the link image IM in which the C003 image is displayed. Then, the display control unit 14 may generate a GIF image after transmitting the C003 image to the user terminal 20 (or simultaneously with the transmission process). For example, the display control unit 14 generates a GIF image configured so that the images are switched in the order of "second place → third place → ... → Nth place → first place" at the frame interval based on the list of the first to Nth contents C2 determined by the image determination unit 131 and the frame interval determined by the frame interval determination unit 132. Then, the display control unit 14 may transmit data of the generated GIF image to the user terminal 20. At this time, the display control unit 14 may also transmit the frame interval to the user terminal 20. Then, the display control unit 14 may instruct the user terminal 20 to change the image displayed as the link image IM from the C003 image to the GIF image when the notified frame interval has elapsed after the C003 image is displayed on the user terminal 20. The above process suppresses the occurrence of a time lag from when the user accesses the top screen SC1 to when the link image IM is displayed, and makes it possible to smoothly draw the link image IM.

[0044] The user log acquiring unit 15 acquires a user log indicating the result of a user's behavior in the content providing service. The user log includes information indicating whether or not a user who accessed the top screen SC1 of the content providing service selected a link image IM, the user ID of the user, information indicating the content C2 displayed as the link image IM (content ID in this embodiment), and the length of the frame interval applied to the link image IM. In addition, the user log corresponding to the case where the link image IM is selected also includes specific information for identifying the content C2 displayed as the link image IM when the link image IM is selected. In this embodiment, as an example, the specific information is information indicating the time from when the link image IM is displayed (i.e., from when the top screen SC1 is displayed) until the link image IM is selected (i.e., the stay time during which the user stayed on the top screen SC1). The user logs for a predetermined period of time related to a plurality of users acquired by the user log acquiring unit 15 are stored in the user log storage unit 10c.

[0045] The model generation unit 16 generates the above-mentioned first model M1 and second model M2 based on a user log indicating the result of the user's action on the setting contents of the link image IM by the link image setting unit 13. In this embodiment, the model generation unit 16 can acquire the user log required for generating the first model M1 and the second model M2 by referring to the user log storage unit 10c. The user log acquisition unit 15 may acquire the user log in a timely manner and store it in the user log storage unit 10c every time the content providing service is used by the user. Then, the model generation unit 16 may update the first model M1 and the second model M2 using the newly acquired user log every time a new user log is stored in the user log storage unit 10c (or at every predetermined fixed period). According to such a process, the first model M1 and the second model M2 can be appropriately updated at any time.

[0046] The model generation unit 16 includes a first model generation unit 161 that generates a first model M1, and a second model generation unit 162 that generates a second model M2.

[0047] As described above, the first model M1 (see Equation 1) is configured to output, for each content a, a first score related to the probability that the link image IM will be selected by the user when the image of the content a is associated with the link image IM (i.e., when an image showing the contents of the content a is displayed to the user as the link image IM) by inputting a user attribute. The first model generation unit 161 generates such a first model M1 by machine learning using information obtained from the above-mentioned user log. More specifically, the first model generation unit 161 generates the first model M1 by machine learning based on information indicating whether the link image IM has been selected by the user, user attributes corresponding to the user ID of the user (in this embodiment, a user attribute vector z corresponding to the user ID stored in the user attribute storage unit 10a), and information indicating the content C2 displayed as the link image IM.

[0048] As described above, in this embodiment, the first model M1 has a parameter θ a (see (Equation 1)). The first model generation unit 161 calculates a reward value r a and the reward value r a and the user attribute vector z of the user, the parameter θ corresponding to the content a a According to such reinforcement learning, in a situation where there is no correct answer data on how to set the link image IM to allow the user to select the link image IM, an index (first score "S_cont a ") can be derived.

[0049] The parameter update method by the above-mentioned reinforcement learning will be described in detail. Here, as an example, an update method using the above-mentioned LinUCB algorithm will be described. First, the first model generation unit 161 updates the parameter θ a and A a is set based on the following (Equation 3) to (Equation 5).

number

[0050] Parameter b a is the parameter θ a is a parameter included in Equation 3. a The initial value of (Equation 3) is "0 d×1 " is a zero vector of the same dimension (d) as the user attribute vector z. (Equation 4) is the parameter A a The initial value of "I d " is the dth order unit matrix. Based on (Equation 5), the parameter θ a As shown in (Equation 3) to (Equation 5), the parameter θ a The initial value of is the zero vector.

[0051] The first model generation unit 161 calculates a parameter A of the corresponding content a for each user action (i.e., for each user log) using the following (Equation 6) and (Equation 7). a and b a Update (learn).

number

[0052] According to (Equation 6), the information that "a user having the user attribute vector z viewed a link image IM associated with an image of the content a (including the case where the link image IM was not selected)" is expressed as the parameter A a In addition, this information is added to the parameter θ a is reflected in.

[0053] According to (Equation 7), the parameter b a is the reward value r a The parameter b is updated based on the user attribute vector z of the user. Note that the term "update" here includes cases where the value does not change before and after the update process. a By updating the parameter θ a Also, the reward value r a and the user attribute vector z of the user (see equation 5).

[0054] Here, the first model generation unit 161 generates a reward value r when a link image IM associated with an image of the content a is selected by a user having a user attribute vector z, the reward value r being larger than the reward value when the link image IM is not selected by the user. a For example, when a link image IM associated with an image of content a is selected, the first model generation unit 161 assigns a reward value r a On the other hand, if a link image IM associated with the image of the content a is not selected, the first model generation unit 161 sets the reward value r a is set to 0. In other words, if a link image IM associated with an image of content a is not selected, the first model generation unit 161 does not grant a reward for that content a.

[0055] By the above-described processing of the first model generating unit 161, θ a The higher the probability that a user (a user with user attribute vector z) will select the link image IM when the image of content a is displayed as the link image IM, the greater the utilization term "θ a T It is learned (updated) so that "z" becomes larger.

[0056] Here, in this embodiment, as described above, the link image IM is associated with a plurality of images of the content C2, and the image displayed as the link image IM changes with the passage of time. In such a case, it is considered that the image of the content C2 displayed at the time when the user selected the link image IM contributed most to inducing such a user's behavior (i.e., selection of the link image IM). On the other hand, it is considered that one or more images of the content C2 displayed as the link image IM from when the link image IM was presented to the user until the user selected the link image IM also contributed to inducing the user's behavior. More specifically, it is considered that the closer the content C2 was displayed as the link image IM to the time when the link image IM was selected by the user, the greater the degree of contribution to inducing the user's behavior. In light of the above, the first model generation unit 161 assigns a reward value r that is larger than the case where the link image IM was not selected by the user to each of the content that was displayed as the link image IM at the time when the link image IM was selected and the content that was displayed as the link image IM before the time when the link image IM was selected. a Then, the reward value r a may be set to a larger value for a content that was displayed as a link image IM closer to the time when the link image IM was selected.

[0057] For example, the first model generation unit 161 calculates the reward value r of each piece of content a based on the following (Equation 8). a may be determined.

number

[0058] In formula 8, "k" indicates the number (display order) of content a displayed as link image IM. The number k of content a displayed nth as link image IM is expressed as "n-1". "i" indicates the number of the content displayed as link image IM when the link image IM was selected by the user. "L" indicates the total number of contents associated with link image IM (i.e., contents included in the GIF image). "Tint" indicates the length of the frame interval applied to link image IM. "ST" indicates the time the user stayed on top screen SC1 (i.e., the time from when top screen SC1 was displayed to when link image IM was selected).

[0059] The first equation (top line) in (Equation 8) applies to content having a number smaller than the number of the content displayed when the link image IM was selected. The second equation (middle line) in (Equation 8) applies to content having a number larger than the number of the content displayed when the link image IM was selected, and which had never been displayed before the link image IM was selected. The third equation (bottom line) in (Equation 8) applies to content having a number larger than the number of the content displayed when the link image IM was selected, and which had been displayed at least once before the link image IM was selected.

[0060] For example, consider the case where "L=9" and the image of the content "k=2" is displayed in the link image IM on the second round and the link image IM is selected. In this case, the reward value r a is calculated as shown in Table 1 below. [Table 1]

[0061] As mentioned above, when a link image IM is selected, the content displayed as the link image IM is given the maximum reward value r aIn addition, a reward value r is assigned to the contents that the user has seen before selecting the link image IM. The ... a By providing this, it is possible to provide each content with an appropriate reward according to its degree of contribution.

[0062] As described above, the second model M2 (see Equation 2) is configured to input a user attribute and output a second score for each frame interval length t (each candidate for a predetermined length t) regarding the probability that the link image IM will be selected by the user when a certain frame interval length t is applied to the link image IM. The second model generation unit 162 generates such a second model M2 by machine learning using information obtained from the above-mentioned user log. More specifically, the second model generation unit 162 generates the second model M2 by machine learning based on information indicating whether the link image IM has been selected by the user, user attributes corresponding to the user ID of the user (in this embodiment, a user attribute vector z corresponding to the user ID stored in the user attribute storage unit 10a), and information indicating the length of the frame interval applied to the link image IM.

[0063] As described above, in this embodiment, the second model M2 has a parameter ρ t (see Equation 2). The second model generation unit 162 calculates a reward value r for the length t of the frame interval applied to the link image IM when the link image IM is selected by the user. t and the reward value r t and the user attribute vector z of the user, the parameter ρ t According to such reinforcement learning, in a situation where there is no correct answer data on how to set the link image IM to allow the user to select the link image IM, an index (second score “S_int tIt is possible to generate a model from which the above (") can be derived.

[0064] The method for updating parameters by the above reinforcement learning will be described in detail. Here, as an example, the update method using the above-described LinUCB algorithm will be described. First, the second model generation unit 162 calculates the parameters ρ t and B t for each length t of the frame interval shown in (Equation 2) based on the following (Equations 9) to (11).

Equation

[0065] The parameter c t is a parameter included in the parameter ρ t . (Equation 9) shows the initial value of the parameter c t . "0 d×1 " in (Equation 9) is a zero vector of the same dimension (d) as the user attribute vector z. (Equation 10) shows the initial value of the parameter B t . "I d " in (Equation 10) is the d-th identity matrix. Based on (Equation 11), the parameter ρ t is calculated. As shown in (Equations 9) to (11), the initial value of the parameter ρ t is a zero vector.

[0066] For each action of the user for one time (that is, for one user log), the second model generation unit 162 updates (learns) the parameter B t and c t for the length t of the corresponding frame interval according to the following (Equations 12) and (13).

Equation

[0067] According to (Equation 12), the information that "a user having a user attribute vector z viewed a link image IM to which the frame interval length t was applied (including the case where the link image IM was not selected)" is expressed as the parameter B t In addition, this information is added to the parameter θ a is reflected in.

[0068] According to (Equation 13), the parameter c t is the reward value r t The parameter c is updated based on the user attribute vector z of the user. Note that the term "update" here includes cases where the value does not change before and after the update process. t By updating, the parameter ρ t Also, the reward value r t and the user attribute vector z of the user (see equation 11).

[0069] Here, the second model generation unit 162 generates a reward value r when a link image IM to which the frame interval length t is applied is selected by a user having a user attribute vector z, the reward value r being larger than the reward value when the link image IM is not selected by the user. t For example, when a link image IM to which a frame interval length t is applied is selected, the second model generation unit 162 assigns a reward value r t On the other hand, if a link image IM to which the frame interval length t is applied is not selected, the second model generation unit 162 sets the reward value r t is set to 0. In other words, if a link image IM to which the length t of the frame interval is applied is not selected, the second model generation unit 162 does not give a reward for that length t.

[0070] By the above-described processing of the second model generating unit 162, ρ t The higher the probability that a user (a user with user attribute vector z) will select a link image IM when the frame interval length t is applied to the link image IM, the greater the utilization term "ρ tT It is learned (updated) so that "z" becomes larger.

[0071] Next, an example of processing by the server 10 will be described with reference to the flowchart shown in FIG.

[0072] First, the request receiving unit 11 receives an access request to the content providing service (top screen SC1) from the user terminal 20 (step S1). Next, the user attribute acquiring unit 12 acquires a user attribute vector z of the user (target user) of the user terminal 20 that is the source of the access request by referring to the user attribute storage unit 10a (step S2).

[0073] Next, the link image setting unit 13 determines a plurality of contents C2 to be associated with the link image IM and the length of the frame interval to be applied to the link image IM (steps S3 to S6).

[0074] Specifically, the image determination unit 131 inputs the user attribute vector z of the target user into the first model M1 ((Formula 1) for each content a) to obtain a first score “S_cont a Then, the image determination unit 131 calculates the first score "S_cont a Based on the first score "S_cont a " determines the top N contents C2 as the images to be associated with the link image IM.

[0075] In addition, the frame interval determination unit 132 inputs the user attribute vector z of the target user to a second model M2 ((Equation 2) for each frame interval length t) to obtain a second score “S_int t" for each length t (step S5). Then, the frame interval determination unit 132 calculates the second score "S_int t ", the frame interval determination unit 132 determines the length t of the frame interval to be applied to the link image IM (step S6). t " is determined as the frame interval to be applied to the link image IM.

[0076] Next, the display control unit 14 presents the top screen SC1 as shown in FIG. 2 to the user (steps S7 to S9). As an example, the display control unit 14 first selects an image to be displayed first as a link image IM (i.e., the image with the first score “S_cont a ") to the user terminal 20 (step S7). As a result, in the user terminal 20, a link image IM in which the image is displayed is generated and presented. After the process of step S7 (or simultaneously with the process of step S7), the display control unit 14 generates a GIF image based on information indicating the plurality of contents C2 determined by the image determination unit 131 (content IDs of the contents C2 in the first to Nth places) and the frame interval determined by the frame interval determination unit 132 (step S8). The GIF image is an animation image configured so that images are switched in the order of "second place → third place → ... → Nth place → first place" at the frame interval. Next, the display control unit 14 transmits the GIF image and information indicating the frame interval to the user terminal 20 (step S9). As a result, when the image (hereinafter referred to as the "initial image") transmitted to the user terminal 20 in step S7 is displayed on the user terminal 20 and the frame interval notified in step S9 has elapsed, the image displayed as the link image IM is changed from the initial image to a GIF image.

[0077] Next, the user log acquisition unit 15 acquires a user log indicating the result of the user's behavior in the content providing service (step S10). That is, the user log acquisition unit 15 acquires a user log which is result information (feedback information) indicating whether or not the user who accessed the top screen SC1 selected the link image IM.

[0078] Next, the model generation unit 16 (the first model generation unit 161 and the second model generation unit 162) updates the first model M1 and the second model M2 based on the user log acquired in step S10 (step S11).

[0079] In the server 10 described above, when there are a plurality of contents that cannot be displayed on one screen (top screen SC1), a link image IM having a link function to a genre detail screen SC2 for displaying a content C2 that is not displayed on the top screen SC1 is displayed on the top screen SC1. In addition, the first model generation unit 161 inputs a user attribute (in this embodiment, a user attribute vector z) and generates a first score "S_cont a A first model M1 that outputs a first score "S_cont a By preferentially associating images with content that has a high "value" with the link image IM, the probability of the link image IM being selected can be improved. As a result, more content can be exposed to the user's eyes, and the conversion rate can be improved.

[0080] Furthermore, the display control unit 14 displays a list of some of the contents C1 associated with each of the multiple genres and a link image IM for each genre on the top screen SC1 (see FIG. 2). Then, the image determination unit 131 determines an image to be associated with the link image IM for each genre. According to the above configuration, by displaying content related to each of the multiple genres on the top screen SC1, it is possible to present content of various genres to the user. Furthermore, by preparing the above-mentioned link image IM for each genre, it is possible to expose more content for each genre to the user's eyes, and therefore it is possible to effectively improve the conversion rate.

[0081] Furthermore, the image determination unit 131 determines each image of the plurality of contents C2 (in this embodiment, N contents C2 with the first scores from 1st to Nth) as an image associated with the link image IM. Then, the display control unit 14 changes the image displayed as the link image IM over time. That is, the link image IM is configured as a GIF image (animation image). According to the above configuration, by changing the image displayed as the link image IM over time, it is possible to bring the plurality of contents C2 into the user's eye through the link image IM on the top screen SC1. By displaying the plurality of contents C2 that may attract the user's attention in sequence on the link image IM, the selection probability of the link image IM can be effectively improved.

[0082] Moreover, the server 10 has a second model generation unit 162 that generates the second model M2, and a frame interval determination unit 132 that determines the frame interval to be applied to the link image IM using the second model M2, and the display control unit 14 changes the image displayed as the link image IM at the constant frame interval determined by the frame interval determination unit 132. According to the above configuration, by applying an appropriate frame interval to the link image IM in accordance with the user attributes, it is possible to effectively improve the selection probability of the link image IM.

[0083] The above embodiment (i.e., the configuration and processing contents of the server 10) may be modified as appropriate. For example, only one image of content may be associated with the link image IM. In other words, the link image IM does not need to be configured so that the content to be displayed is switched at a fixed frame interval. In other words, the link image IM may be configured with only an image of one specific content. In this case, the frame interval determination unit 132, the second model M2, and the second model generation unit 162 may be omitted. Also, in this case, the image determination unit 131 determines the first score "S_cont a The image with the largest number of links may be determined as the image to be associated with the link image IM.

[0084] In addition, the machine learning algorithm executed by the first model generation unit 161 to generate the first model M1 is not limited to the above-mentioned LinUCB algorithm (bandit algorithm). a " (i.e., a score according to the probability that the link image IM will be selected by the user), and an algorithm different from the algorithm (parameter update formula) described in the above embodiment may be used. For example, in the above embodiment, the LinUCB algorithm based on the so-called UCB strategy among bandit algorithms is used, but an algorithm of a strategy other than the above (e.g., ε-greedy strategy, Thompson Sampling strategy, etc.) may be used. Also, a reinforcement learning algorithm other than a bandit algorithm may be used. The same applies to the second model generation unit 162 and the second model M2.

[0085] The block diagrams used to explain the above embodiments show functional blocks. These functional blocks (components) are realized by any combination of at least one of hardware and software. The method of realizing each functional block is not particularly limited. That is, each functional block may be realized by using one device that is physically or logically combined, or may be realized by using two or more devices that are physically or logically separated and directly or indirectly connected (for example, by wire, wirelessly, etc.). The functional blocks may be realized by combining the one device or the multiple devices with software.

[0086] Functions include, but are not limited to, judging, determining, assessing, calculating, processing, deriving, investigating, searching, verifying, receiving, sending, outputting, accessing, resolving, selecting, choosing, establishing, comparing, anticipating, expecting, regarding, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, and assigning.

[0087] For example, the server 10 according to an embodiment of the present disclosure may function as a computer that performs the information processing method of the present disclosure. Fig. 5 is a diagram showing an example of a hardware configuration of the server 10 according to an embodiment of the present disclosure. The above-mentioned server 10 may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, and the like.

[0088] In the following description, the term "apparatus" may be replaced with a circuit, a device, a unit, etc. The hardware configuration of the server 10 may be configured to include one or more of the apparatuses shown in FIG. 5, or may be configured to exclude some of the apparatuses.

[0089] Each function of the server 10 is realized by loading a specific software (program) onto hardware such as a processor 1001 and a memory 1002, so that the processor 1001 performs calculations, controls communications via a communication device 1004, and controls at least one of reading and writing of data in the memory 1002 and the storage 1003.

[0090] The processor 1001 controls the entire computer by running, for example, an operating system, and may be configured with a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic unit, a register, and the like.

[0091] Moreover, the processor 1001 reads out a program (program code), a software module, data, etc. from at least one of the storage 1003 and the communication device 1004 into the memory 1002, and executes various processes according to the read out programs. As the program, a program that causes a computer to execute at least a part of the operations described in the above-mentioned embodiment is used. For example, the link image setting unit 13 may be realized by a control program stored in the memory 1002 and running on the processor 1001, and other functional blocks may be similarly realized. Although the above-mentioned various processes have been described as being executed by one processor 1001, they may be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The program may be transmitted from a network via a telecommunication line.

[0092] The memory 1002 is a computer-readable recording medium, and may be configured by at least one of, for example, a Read Only Memory (ROM), an Erasable Programmable ROM (EPROM), an Electrically Erasable Programmable ROM (EEPROM), a Random Access Memory (RAM), etc. The memory 1002 may be called a register, a cache, a main memory (primary storage device), etc. The memory 1002 can store a program (program code), software modules, etc. that are executable to implement a communication control method according to an embodiment of the present disclosure.

[0093] Storage 1003 is a computer-readable recording medium, and may be composed of at least one of, for example, an optical disk such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital versatile disk, a Blu-ray (registered trademark) disk), a smart card, a flash memory (e.g., a card, a stick, a key drive), a floppy (registered trademark) disk, a magnetic strip, and the like. Storage 1003 may be referred to as an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, a server, or other suitable medium including at least one of memory 1002 and storage 1003.

[0094] The communication device 1004 is hardware (transmission / reception device) for performing communication between computers via at least one of a wired network and a wireless network, and is also called, for example, a network device, a network controller, a network card, or a communication module.

[0095] The input device 1005 is an input device (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.) that accepts input from the outside. The output device 1006 is an output device (e.g., a display, a speaker, an LED lamp, etc.) that performs output to the outside. The input device 1005 and the output device 1006 may be integrated into one device (e.g., a touch panel).

[0096] In addition, each device such as the processor 1001 and the memory 1002 is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or may be configured using different buses between each device.

[0097] The server 10 may also be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), a field programmable gate array (FPGA), etc., and some or all of the functional blocks may be realized by the hardware. For example, the processor 1001 may be implemented using at least one of these pieces of hardware.

[0098] Although the present embodiment has been described in detail above, it is clear to those skilled in the art that the present embodiment is not limited to the embodiment described in this specification. The present embodiment can be implemented as a modified or altered form without departing from the spirit and scope of the present invention defined by the description of the claims. Therefore, the description in this specification is intended to be illustrative and does not have any limiting meaning on the present embodiment.

[0099] The order of the steps, sequences, flow charts, etc. of each aspect / embodiment described in this disclosure may be changed unless inconsistent. For example, the methods described in this disclosure present elements of various steps using an example order, and are not limited to the particular order presented.

[0100] The input and output information may be stored in a specific location (e.g., memory) or may be managed using a management table. The input and output information may be overwritten, updated, or added to. The output information may be deleted. The input information may be transmitted to another device.

[0101] The determination may be based on a value represented by a single bit (0 or 1), a Boolean value (true or false), or a numerical comparison (e.g., comparison with a predetermined value).

[0102] Each aspect / embodiment described in this disclosure may be used alone, in combination, or switched according to execution. In addition, notification of specific information (e.g., notification that "X is the case") is not limited to being done explicitly, but may be done implicitly (e.g., not notifying the specific information).

[0103] Software shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.

[0104] Additionally, software, instructions, information, etc. may be transmitted or received over a transmission medium. For example, if the software is transmitted from a website, server, or other remote source using wired and / or wireless technologies (such as coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL)), and / or wireless technologies (such as infrared, microwave, etc.), then these wired and / or wireless technologies are included within the definition of transmission media.

[0105] The information, signals, etc. described in this disclosure may be represented using any of a variety of different technologies. For example, the data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0106] Furthermore, the information, parameters, etc. described in this disclosure may be expressed using absolute values, may be expressed using relative values ​​from a predetermined value, or may be expressed using other corresponding information.

[0107] The names used for the above-mentioned parameters are not limiting in any way. Furthermore, the formulas etc. using these parameters may differ from those explicitly disclosed in this disclosure. The various information elements may be identified by any suitable names, and the various names assigned to these various information elements are not limiting in any way.

[0108] As used in this disclosure, the phrase "based on" does not mean "based only on," unless expressly stated otherwise. In other words, the phrase "based on" means both "based only on" and "based at least on."

[0109] Any reference to an element using a designation such as "first," "second," etc., used in this disclosure does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient method of distinguishing between two or more elements. Thus, a reference to a first and a second element does not imply that only two elements may be employed or that the first element must precede the second element in some way.

[0110] When the terms "include," "including," and variations thereof are used in this disclosure, these terms are intended to be inclusive, similar to the term "comprising." Further, when used in this disclosure, the term "or" is not intended to be an exclusive or.

[0111] In this disclosure, where articles have been added due to translation, such as a, an, and the in English, the disclosure may include that the nouns following these articles are in the plural form.

[0112] In the present disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "coupled" may also be interpreted in the same way as "different." [Explanation of symbols]

[0113] 10...server (information processing device), 11...request receiving unit, 12...user attribute acquisition unit, 13...link image setting unit, 14...display control unit, 15...user log acquisition unit, 16...model generation unit, 20...user terminal, 21...display unit, 131...image determination unit, 132...frame interval determination unit, 161...first model generation unit, 162...second model generation unit, C1...content, C2...content (hidden content), IM...link image, M1...first model, M2...second model, SC1...top screen (first screen), SC2...genre details screen (second screen).

Claims

1. a display control unit that displays, in a first screen presented to a user, a list of some of the plurality of contents, and a link image in which a link to a second screen in which a list of non-displayed contents not displayed in the first screen is displayed and one or more images of the non-displayed contents are associated; a first model generation unit that generates a first model by machine learning based on information indicating whether the link image has been selected by a user, the user attributes of the user, and information indicating the content displayed as the link image, the first model being configured to output, for each piece of content, a first score corresponding to a user attribute relating to a probability that the link image will be selected by the user when the image of the content is associated with the link image; a user attribute acquisition unit that acquires user attributes of a target user to whom the first screen is presented; an image determination unit that obtains the first score for each piece of content by inputting the user attributes of the target user obtained by the user attribute acquisition unit into the first model, and determines, preferentially, an image of the content having a high first score as an image to be associated with the link image in the first screen presented to the target user; An information processing device comprising:

2. the first model includes parameters for each content; The first model generation unit When the link image is selected by the user, a reward value that is greater than a reward value in a case where the link image is not selected by the user is given to the content that was displayed as the link image; The information processing apparatus according to claim 1 , further comprising: updating the parameter corresponding to the content based on the reward value and a user attribute of the user.

3. the display control unit displays, on the first screen, a list of some of the contents associated with each of a plurality of genres related to the contents and the link image for each of the genres; The information processing apparatus according to claim 1 , wherein the image determination unit determines an image to be associated with the link image for each of the genres.

4. the image determination unit determines an image of each of the plurality of non-display contents as an image to be associated with the link image; 4. The information processing device according to claim 1, wherein the display control unit changes the image displayed as the link image in accordance with the passage of time.

5. the display control unit changes the image displayed as the link image at regular frame intervals; a second model generation unit that generates a second model by machine learning based on information indicating whether the link image has been selected by a user, the user attributes of the user, and information indicating the length of the applied frame interval, the second model being configured to output a second score corresponding to a user attribute relating to the probability that the link image will be selected by the user for each length of the frame interval; 5. The information processing device of claim 4, further comprising a frame interval determination unit that obtains the second score for each frame interval length by inputting user attributes of the target user into the second model, and preferentially determines the frame interval length having a higher second score as the frame interval length to be applied to the link image in the first screen presented to the target user.

6. the second model includes parameters for each length of the frame interval; The second model generation unit When the link image is selected by the user, a reward value is assigned to the length of the frame interval applied to the link image that is greater than a reward value assigned to a case in which the link image is not selected by the user; The information processing apparatus according to claim 5 , further comprising: updating the parameter corresponding to the length of the frame interval based on the reward value and a user attribute of the user.

7. the first model includes parameters for each content; The first model generation unit a reward value greater than a reward value obtained when the link image is not selected by the user is given to each of the content displayed as the link image at the time when the link image is selected and the content displayed as the link image before the link image is selected; generating the first model by reinforcement learning that updates the parameters corresponding to each piece of content based on the reward value and a user attribute of the user; The reward value is set to a larger value for the content that was displayed as the link image closer to the time point when the link image was selected.

7. The information processing device according to claim 5 or 6.

8. the display control unit causes the first screen and the second screen to be displayed on a display unit provided in a user terminal different from the information processing device; The information processing device according to any one of claims 4 to 7, wherein the display control unit transmits to the user terminal an image to be first displayed as the link image among the images of each of the plurality of hidden contents determined by the image determination unit, and then transmits to the user terminal an animation image configured so that each of the images of the plurality of hidden contents changes over time.

Images