JP2024028005A - Extraction device, extraction method and extraction program - Google Patents

Abstract

To extract users who can increase the distribution effect of a content.SOLUTION: An extracting apparatus of the present application has an acquiring unit, a specifying unit, a score calculating unit, an operation rate calculating unit, and an extracting unit. The acquiring unit acquires an action history of a user. The identifying unit identifies features of a content to be distributed to the user. The score calculating unit calculates, for each user, a similarity score indicating the degree of similarity of the action history to features of the content based on the action history. The operation rate calculation unit calculates, for each user, an operation rate indicating a ratio of operations performed by the user on the content distributed to the user. The extraction unit extracts users with a high operation rate from a plurality of users with a high similarity score.SELECTED DRAWING: Figure 2

Description

The present invention relates to an extraction device, an extraction method, and an extraction program.

In recent years, information distribution via the Internet has become popular. As an example of such information distribution, a click rate such as CTR (Click Through Rate) is estimated based on the similarity of advertisements to be distributed and information on users to whom advertisements are distributed. 2. Description of the Related Art There is a known technology for selecting advertisements to be distributed according to click rate.

For example, a technique is known in which the CTR is predicted based on the constituent elements of advertisements to be distributed or the similarity between advertisements, and the predicted CTR is used to select advertisements to be distributed. In addition, the technology calculates the virtual CTR based on the attributes of the user who selected the advertisement, and selects the advertisement to be distributed according to the attributes of the user to whom the advertisement is distributed and the calculated virtual CTR. Are known.

Japanese Patent Application Publication No. 2014-006684 Japanese Patent Application Publication No. 2014-182473

However, even if advertisements are distributed based on the attributes of the users to whom the advertisements are distributed, some users have low click-through rates, so Even if there is a correlation between the two, distributing advertisements to such users is likely to lead to a decrease in click rate. In other words, users who do not perform operations such as selection on distributed advertisements are unlikely to perform operations on advertisements even if advertisements are distributed, so users who do not perform operations on advertisements On the other hand, even if advertisements are distributed, the advertising effect tends to be low relative to the number of distributed advertisements. As mentioned above, users who do not perform operations such as clicking on distributed content such as advertisements are unlikely to perform operations such as selecting and viewing content even if the content is distributed. Even if content is distributed to users who do not perform operations on the content, the effectiveness of the distribution is likely to be reduced.

The present application has been made in view of the above, and aims to provide an extraction device, an extraction method, and an extraction program that can extract users who can enhance the effectiveness of content distribution.

The extraction device according to the present application includes an acquisition unit that acquires the behavior history of a user, a specification unit that identifies the characteristics of content to be distributed to the user, and a determination unit that determines the similarity of the behavior history to the characteristics of the content. a score calculation unit that calculates a similarity score for each user based on the action history; and an operation rate that shows the rate at which the user performed an operation on the content distributed to the user. The present invention is characterized in that it includes an operation rate calculation unit that calculates the operation rate for each user, and an extraction unit that extracts the user with the high operation rate from among the plurality of users who have the high similarity score.

According to one aspect of the embodiment, it is possible to extract users who can enhance the content distribution effect.

FIG. 1 is a diagram illustrating an example of an extraction process executed by an information providing apparatus according to an embodiment. FIG. 2 is a diagram illustrating a configuration example of an information providing apparatus according to an embodiment. FIG. 3 is a schematic diagram of a user ID list. FIG. 4 is an explanatory diagram when extracting high operation rate IDs from the user ID list shown in FIG. 3. FIG. 5 is a flowchart illustrating an example of the flow of extraction processing executed by the information providing apparatus according to the embodiment. FIG. 6 is an explanatory diagram when extracting IDs of users with high advertisement click rate and advertisement conversion rate. FIG. 7 is an explanatory diagram when extracting IDs of users with high index 1 and index 2 of operation rate. FIG. 8 is a diagram showing an example of the hardware configuration.

EMBODIMENT OF THE INVENTION Below, the extraction apparatus, the extraction method, and the form (henceforth described as an "embodiment") for implementing the extraction program which concern on this application are demonstrated in detail, referring drawings. Note that the extraction device, extraction method, and extraction program according to the present application are not limited to this embodiment. Further, in each of the embodiments below, the same parts are given the same reference numerals, and redundant explanations will be omitted.

[Embodiment]
[1. Regarding the processing displayed by the information providing device]
First, an example of an extraction process executed by an information providing apparatus 10, which is an example of an extraction apparatus, will be described using FIG. FIG. 1 is a diagram illustrating an example of an extraction process executed by an information providing apparatus according to an embodiment.

In the following explanation, the extraction process executed by the information providing device 10 will be based on the user's behavior history, the distribution history of advertising-related content, and the operation history for distributed content (hereinafter referred to as "log"). , an example of the process of calculating the similarity between the user's action history and the characteristics of the content and extracting the users who have a high operation rate of operating the content will be described, but the embodiment is not limited to this. do not have. The information providing device 10 may extract users who have a high content manipulation rate from the distribution history of arbitrary content such as news feeds and images and logs for the content.

Furthermore, in the following description, advertisement is a concept that includes not only commercial or non-commercial advertisements, but also volunteer recruitment, public advertisements, notifications to the public, and other arbitrary content. In addition, advertisements are not only content that includes so-called advertising-related information, but also information that arouses the user's interest and is included in information about or related to an arbitrary target (for example, a landing page, etc.) Images, moving images, text, graphics, symbols, hyperlinks, and other arbitrary content may be included along with the text as long as the information is widely disseminated.

[1-1. Example of information providing device]
The information providing device 10 is an information processing device that can communicate with a log server 100, a distribution server 200, and a terminal device 300 used by a user via a predetermined network N such as the Internet (see FIG. 2, for example). For example, it is realized by a server device, a cloud system, etc. Note that the information providing device 10 may be able to communicate with an arbitrary number of log servers 100 and distribution servers 200 via the network N, and may be able to communicate with various external servers (not shown), for example. good. Further, although only one terminal device 300 is illustrated in FIG. 1, the terminal device 300 is capable of communicating with a large number of terminal devices 300 via the network N.

The log server 100 is a server device that manages logs, and is realized by, for example, a server device, a cloud system, or the like. For example, the log server 100 can store websites accessed by the user using the terminal device 300, web content distributed to the terminal device 300 used by the user, advertisement content distributed to the terminal device 300 used by the user, It acquires and manages various log data associated with content distribution, such as the content selected by the user using the terminal device 300 from among the distributed content, the demographic attributes and psychographic attributes of the user, and the like. Note that the functions performed by the log server 100 are the same as those of well-known log servers, and detailed description thereof will be omitted.

The distribution server 200 is an information processing device that distributes content to the terminal device 300 used by the user, and is realized by, for example, a server device, a cloud system, or the like. For example, the distribution server 200 selects content that is likely to be selected and operated by the user based on various attributes and behavior history of the user who has accessed the web content, and distributes the selected content. To give a more specific example, the distribution server 200 lists advertisements based on the user's behavior history, and distributes advertisements selected as a result of the listing.

The terminal device 300 is a smart device such as a desktop PC (Personal Computer), a notebook PC, a smartphone, or a tablet, and is a device that can communicate with any server device via the network N. The terminal device 300 includes an input device through which a user performs input operations, and an output device through which an output is output to the user. The output device is a display unit that displays information to the user. As the input device, for example, a mouse or a keyboard may be used, or a touch panel integrated with a display unit may be used.

Here, when displaying web content such as a web page by a user's operation, the terminal device 300 transmits a web content distribution request to a web server that distributes the web content. In such a case, upon receiving the distribution of the web content, the terminal device 300 transmits a distribution request for an advertisement placed in the web content to the distribution server 200.

In such a case, the distribution server 200 distributes the advertisement image to the terminal device 300. Then, the terminal device 300 arranges and displays the advertisement image received from the distribution server 200 within the web page. In addition, when an operation to select an advertisement image is performed, such as when the user clicks the cursor on the advertisement image or when the user taps the advertisement image, the terminal device 300 displays the landing page corresponding to the advertisement image. While displaying the page, a notification to the effect that the advertisement image has been selected is transmitted to the distribution server 200. Further, operations performed by the user on the terminal device 300 as described above are acquired by the log server 100 as log data.

[1-2. Example of extraction process]
Here, in order to deliver content that is likely to be selected or otherwise performed by the user, we compare the attributes and behavior history of the user with the features of the content, and select features that are highly relevant to the user's behavior history. A possible mode is to distribute the content that the user has to the user. However, some users do not perform many operations on the distributed content. Such users who do not perform operations on content will not perform operations such as selection on the content even if content that is highly relevant to their behavior history is distributed, so the effectiveness of content distribution will be reduced. becomes low. Therefore, the information providing apparatus 10 according to the present embodiment executes the following extraction process.

[1-3. Regarding user behavior history]
First, the information providing device 10 acquires the user's behavior history. In this case, the user's action history includes, for example, the search keyword when the user performed a search on the network, the destination domain when the user performed an operation based on the search results, and the domain to which the user changed. Examples include the flow from the transition source domain to the transition destination domain when the transition occurs. The user's action history acquired in this manner is linked to an ID that is user identification information set for each user.

[1-4. Regarding identification of content characteristics]
Furthermore, the information providing device 10 specifies the characteristics of the content to be distributed to the user. The characteristics of content include, for example, the attributes of the information shown by the content, such as the attributes of products, services, and event information shown by the content, and whether the content arouses interest in users or encourages purchase. Examples include attributes of the content itself, such as whether the content is present or not.

[1-5. Regarding calculation of similarity score]
Further, the information providing device 10 calculates a similarity score for each user based on the user's action history, which indicates the degree of similarity of the user's action history to the content feature. The similarity score is a numerical representation of the similarity or relationship between the content features and the user's behavior history. It is designed to allow comparisons between users depending on the situation. The calculated similarity score is assigned to the user's ID. Furthermore, the similarity score is calculated for each content feature with respect to the user's behavior history. Therefore, the similarity score assigned to the user ID is assigned a score corresponding to each feature of the content distributed to the user.

[1-6. Regarding calculation of operation rate]
Further, the information providing device 10 calculates, for each user, an operation rate that indicates the rate at which users perform operations on content distributed to users. An example of the user operation rate for the content distributed to the user is the click rate, that is, the click through rate (CTR), which is the rate at which the users select the distributed content. Other examples of the user operation rate for content include the rate of purchasing operations based on distributed content, the rate of making reservations for restaurants, accommodation facilities, events, etc. When calculating the user's operation rate for the content distributed to the user, the operation rate is calculated for each operation performed by the user on the content. As for the calculated operation rate, the operation rate for each operation performed by the user on the content is assigned to the user's ID.

[1-7. Regarding extraction of users with high operation rate]
Furthermore, the information providing device 10 extracts a user with a high operation rate from among a plurality of users with a high similarity score. For example, when extracting a user who has a high operation rate for a given operation on a given piece of content, select the user who has a high similarity score for the feature of the content given to the user ID. Extract users with high arbitrary operation rate. In order to achieve this, the information providing device 10 calculates the similarity score for each content feature assigned to the IDs of multiple users with respect to the similarity score of the user's behavior history with respect to the content feature. List them in order. That is, a list of IDs of a plurality of users is created in which the similarity scores assigned to the IDs are arranged in descending order of the scores. When extracting a user with a high operation rate from among multiple users with high similarity scores, the operation Extract users with high rate.

As a result, when distributing content such as advertisements to multiple users, it is possible to extract users who are likely to be interested in the content. It is possible to extract users who can increase the effectiveness of content distribution, such as by viewing the content or purchasing the product. For example, if the content is an advertisement, the advertising effect can be enhanced as the content distribution effect.

[1-8. Regarding an example of extraction processing]
Next, an example of the extraction process executed by the information providing device 10 will be described using FIG. 1. First, the information providing device 10 acquires a user's behavior history from log data managed by the log server 100 (step S1). That is, the information providing device 10 uses the search keyword when the user searches on the network using the terminal device 300, the transition destination domain and the like when the user operates the terminal device 300 based on the search results. , the flow from the transition source domain to the transition destination domain when the domain is transitioned, etc. is acquired from the log server 100 as the user's action history. The information providing device 10 that acquires the user's behavior history in this manner acquires the behavior history of a large number of users included in the log data.

Furthermore, the information providing device 10 specifies the characteristics of the content to be distributed to the user (step S2). The information providing device 10 identifies characteristics of the content, such as fields and types of information included in the content scheduled to be distributed from the distribution server 200, based on the content. If the content to be distributed to users is an advertisement, the characteristics of the advertisement, such as the field of products or services displayed in the advertisement, are specified.

The information providing device 10 that has acquired the user's behavior history and identified the content features calculates and lists the similarity score of the user's behavior history with respect to the content features (step S3). A similarity score is calculated for each feature of content. Further, a similarity score is calculated for each user and assigned to the user's ID. Therefore, a similarity score for each content feature is assigned to the user ID for each content feature. Listing of similarity scores is performed by listing similarity scores for each content feature assigned to each user's ID for each content feature.

In other words, each user's ID is assigned multiple similarity scores for each content feature, but among these similarity scores, the similarity score for the same content feature in each user's ID is A list is created as an ID list 50 that is associated with IDs. The ID list 50 in which each user's ID and similarity score are associated is created, for example, in descending order of similarity score. This makes it possible to determine the relative degree of similarity of the user's behavior history with respect to the feature of any content by comparing the similarity scores on the ID list 50.

After creating the ID list 50, the information providing device 10 calculates the user's operation rate for the distributed content (step S4). The information providing device 10 calculates a user's operation rate for each content feature based on log data managed by the log server 100. In other words, since the ID list 50 is created for each content feature, by calculating the user's operation rate for each content feature, content that has the feature corresponding to the similarity score shown in the ID list 50 is calculated. Calculate the operation rate. In this case, the user's operation rate for the content includes, for example, when the content is an advertisement, the click rate, or CTR (Click Through Rate), which is the rate at which the user selects the distributed advertisement. When calculating the click rate of an advertisement as a user's interaction rate with distributed content, the click rate is calculated for each user. Therefore, for each ID listed in the ID list 50, the click rate of an advertisement having characteristics corresponding to the ID list 50 is calculated.

After calculating the user's operation rate for the content, the information providing device 10 extracts users with a high operation rate (step S5). Specifically, the information providing device 10 extracts the IDs of users with high similarity scores and high operation rates from the ID list 50. For example, thresholds are set for each of the similarity score and operation rate in the ID list 50, and the similarity score is equal to or greater than the threshold set for the similarity score, and the operation rate is equal to or greater than the threshold set for the operation rate. Users with a high operation rate are extracted. In other words, if Score1 is set as a threshold for the similarity score in the ID list 50 and CTR1 is set for the click rate, which is the operation rate, then if the similarity score is Score1 or more and the click rate is CTR1 or more, then The ID of the user is extracted as the high operation rate ID55, which is the ID of the user with a high similarity score and a high click rate.

After extracting the high operation rate ID 55, the list of extracted user IDs is provided to the distribution server 200 (step S6). In other words, since the high operation rate ID55 includes the IDs of multiple users who have a high similarity score and a high click rate, that is, a high operation rate, the information providing device 10 uses the extracted high operation rate ID55. By providing the list to the distribution server 200, IDs of a plurality of users with high similarity scores and high operation rates are provided.

The distribution server 200 provided with the high operation rate ID 55 distributes the content to the extracted users (step S7). In other words, the ID of the user extracted as the high operation rate ID 55 is an ID that has a high interest in the content and is likely to perform an operation such as clicking and making a selection. For this reason, for example, if the content is an advertisement, a user who uses the ID extracted as high operation rate ID 55 has a high interest in the content of the advertisement, and also performs operations such as clicking and viewing the advertisement. are more likely to do so. Therefore, by distributing content having the characteristics used in creating the ID list 50 from which the high operation rate ID 55 was extracted to the terminal device 300 that is used using the ID extracted as the high operation rate ID 55. , it is possible to increase the possibility that a user will perform an operation such as a click on the distributed content.

[2. Configuration of information providing device]
An example of the functional configuration of the information providing device 10 described above will be described below. FIG. 2 is a diagram illustrating a configuration example of an information providing apparatus according to an embodiment. As shown in FIG. 2, the information providing device 10 includes a communication section 20, a storage section 30, and a control section 40.

The communication unit 20 is realized by, for example, a NIC (Network Interface Card). The communication unit 20 is connected to the network N by wire or wirelessly, and transmits and receives information between the log server 100, the distribution server 200, and the terminal device 300.

The storage unit 30 is realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. The storage unit 30 also stores a user database 31 and a content database 32.

Various types of information regarding users are registered in the user database 31. In the user database 31, for example, information such as "user ID", "user characteristic information", "access destination", "access date and time", "distribution content", "search keyword", etc. is registered.

Here, the "user ID" is an identifier that identifies a user. Furthermore, "user characteristic information" refers to various user characteristics used in content listing, such as user demographic attributes and psychographic attributes. Further, the "access destination" is information indicating the access destination, such as a website, accessed by the user indicated by the associated "user ID." Furthermore, the "access date and time" is information indicating the date and time when the user indicated by the associated "user ID" accessed the website etc. indicated by the associated "access destination". Furthermore, "distributed content" is the content ID of the content distributed when the user indicated by the associated "user ID" accesses the website etc. indicated by the associated "access destination". . Moreover, the "search keyword" is the keyword used when the user performed a search on the website.

Various types of information related to content are registered in the content database 32. In the content database 32, for example, information such as "content ID", "content type", "content content", "content keyword", etc. is registered.

Here, the "content ID" is an identifier that identifies content. Further, "content type" is information indicating the type of content, such as advertisement or news. In addition, "content content" is information about the content displayed by the content. For example, if the content is an advertisement, it is information about the type of product displayed by the advertisement, and if the content is news, it is information about the content displayed by the content. This is information about the genre of news. Furthermore, the "keyword of content" is information about keywords that represent the content or genre of the content.

The control unit 40 is a controller, and for example, various programs stored in a storage device inside the information providing device 10 are stored in a RAM or the like by a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). This is achieved by executing this as a work area. Further, the control unit 40 is a controller, and may be realized by, for example, an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

The control unit 40 includes an acquisition unit 41 , a specification unit 42 , a score calculation unit 43 , an operation rate calculation unit 44 , and an extraction unit 45 . The acquisition unit 41 acquires various types of information regarding users and various types of information regarding content from the log server 100. At that time, the acquisition unit 41 acquires the user's characteristic information, access destination, access date and time, distributed content, search keywords, etc. for each user ID, and uses the acquired information about the user with the user ID. It is registered in the user database 31 in association with. Thereby, the acquisition unit 41 acquires the user's action history. In other words, the acquisition unit 41 acquires search keywords, access destinations, access dates and times, distributed contents, etc. when the user searches, which are managed by the log server 100, as the user's action history, and acquires the user's behavior history. It is registered in the user database 31 in association with the ID.

Further, the acquisition unit 41 acquires the content type, content content, content keyword, etc. for each content ID, and registers the information regarding the acquired content in the content database 32 in association with the ID.

The specifying unit 42 specifies the characteristics of content to be distributed to users. More specifically, the identifying unit 42 identifies characteristics of the content, such as fields and types of information included in the content scheduled to be distributed from the distribution server 200, based on the content. For example, the specifying unit 42 refers to the content database 32 and determines the attributes of information indicated by the content, such as the attributes of products, services, and event information indicated by the content, and whether the content arouses interest in the user. The attributes of the content itself, such as whether it is available or encourages purchase, are identified as characteristics of the content.

The score calculation unit 43 calculates the similarity score of the user's behavior history acquired by the acquisition unit 41 with respect to the content feature specified by the identification unit 42 for each user based on the user's behavior history. At this time, the score calculation unit 43 calculates the similarity score of the user's action history for each feature of the content, and assigns the calculated similarity score to each user's ID. Furthermore, the score calculation unit 43 creates a list of user IDs based on the calculated similarity score.

FIG. 3 is a schematic diagram of a user ID list. The user ID list 50 is created by creating a list of similarity scores for each content feature, which are assigned to each user's ID, for each content feature. An ID list 50 in which each user's ID and similarity score are associated with each other for each content feature is created in descending order of similarity score. This makes it possible to compare the relative degree of similarity of each user's behavior history with respect to the characteristics of a given content, and for example, to extract users with relatively high similarity scores. can do.

The operation rate calculating unit 44 calculates, for each user, the user's operation rate for the content distributed to the user. For example, the operation rate calculation unit 44 refers to the user database 31 and calculates the click rate (CTR), which is the rate at which users select content distributed to users, and the purchase rate based on the distributed content. The rate of operations and the rate of reservations for restaurants, accommodation facilities, events, etc. are calculated as the user's operation rate for the content. At this time, the operation rate calculation unit 44 calculates the operation rate when selecting content by clicking, the operation rate when purchasing a product based on the content, etc., to calculate the operation rate for the content. Calculated for each type. Furthermore, the operation rate calculation unit 44 calculates the user's operation rate for the content for each feature of the content. The calculated operation rate is assigned to the ID of the user in association with the characteristics of the content and the type of operation on the content, and is registered in the user database 31.

The extraction unit 45 extracts a user who has a high interaction rate with the distributed content from among the multiple users who have a high similarity score. Specifically, the extraction unit 45 refers to the user ID list 50 created in descending order of similarity score, and extracts from the ID list 50 the users who have a high interaction rate with the distributed content.

FIG. 4 is an explanatory diagram when extracting high operation rate IDs from the user ID list shown in FIG. 3. Since the ID list 50 is created for each feature of the content, the extraction unit 45 selects the distributed user from among the ID list 50 created corresponding to the feature of the content from which users with high operation rates are to be extracted. IDs of users who have a high rate of operation with respect to content are extracted. For example, if the operation rate for the content is the click rate (CTR) for the advertisement compared to the content, the extraction unit 45 selects the click rate (CTR) for the advertisement from the ID list 50 created based on the similarity score for the advertisement. Extract the IDs of users with high rates.

Specifically, the extraction unit 45 sets thresholds for each of the similarity score and the operation rate, and the similarity score is equal to or greater than the threshold set for the similarity score, and the operation rate is equal to or greater than the operation rate. Users whose operation rate is greater than or equal to the threshold value set for the user are extracted as users with a high operation rate. This suppresses the extraction of users with low similarity scores or users with low operation rates when extracting users with high operation rates. For example, Score1 is set as a threshold for the similarity score in the ID list 50, and CTR1 is set as a threshold for the click rate, which is the operation rate. In this way, with respect to the threshold values set for the similarity score and the click rate, the extraction unit 45 determines that if the CTR given to the user's ID is CTR≧CTR1, the similarity given to the ID is The IDs of a plurality of users whose scores satisfy similarity score≧Score1 are extracted as high operation rate ID55, which is the ID of a user who has a high similarity score and a high click rate.

Furthermore, the extraction unit 45 can also extract a plurality of users who can bring the average operation rate, which is the average operation rate of a plurality of users, closer to the target operation rate, which is the target operation rate. ing. When extracting multiple users that can bring the average operation rate of multiple users closer to the target operation rate, by changing the operation rate threshold, the average operation rate of the multiple users to be extracted can be reduced. Get closer to the target operation rate. For example, when setting the target CTR, which is the target click rate, as the target operation rate, by changing CTR1, which is the click rate threshold, the average CTR, which is the average operation rate of multiple extracted users, can be set. , get closer to the target CTR.

Specifically, when the number of user IDs included in the ID list 50 is N and the target CTR is CTR _GOAL , the CTR of each of the N IDs is calculated as shown in equation (1) below. CTR1 is changed so that the average CTR approaches the CTR _GOAL .

Furthermore, when bringing the average operation rate of a plurality of users closer to the target operation rate, the extraction unit 45 also changes the threshold of the similarity score. In other words, the extraction unit 45 brings the average CTR closer to the CTR _GOAL by also changing Score1, which is the threshold of the similarity score. The extraction unit 45 brings the average operation rate of a plurality of users closer to the target operation rate by repeatedly changing the operation rate threshold and the similarity score threshold as described above.

Note that when extracting users with a high interaction rate with respect to content, the extraction unit 45 calculates that the value calculated based on the multiplication of the similarity score and the interaction rate is based on the multiplication of the similarity score and the interaction rate. It is preferable to extract users with a high operation rate from among the users whose operation rate is equal to or higher than the lower limit value A set for the value calculated based on the calculated value.

In other words, if multiple values calculated based on the multiplication of multiple similarity scores and multiple CTRs are expressed on the ID list 50, a curve connecting the portion with CTR 1 and the portion with Score 1 in high operation rate ID 55 56 (see FIG. 4). Therefore, when extracting users with a high operation rate for content, an ID with a higher CTR than the curved portion 56 and a higher similarity score than the curved portion 56 is extracted as the high operation rate ID 55. In this way, when extracting the high operation rate ID 55 using the curve part 56, by setting the lower limit value A for the value calculated based on the multiplication of the similarity score and the operation rate, the curve part 56 can be extracted. 56, it is possible to adjust the IDs included in the high operation rate ID55.

In addition, when extracting the high operation rate ID55 using the curve part 56, in order to bring the average operation rate of multiple users closer to the target operation rate, CTR1, which is the operation rate threshold, or the similarity score threshold In addition to changing Score1, the lower limit value A that is set for the value calculated based on the multiplication of the similarity score and the operation rate is also changed. In other words, when the curve part 56 indicates similarity score x CTR, the relationship between similarity score, CTR, and lower limit value A is similarity score x CTR ≧A, so lower limit value A is changed. By doing so, the extracted similarity score and CTR can be changed. Thereby, the average operation rate can be changed, so the average operation rate can be changed to bring the average operation rate closer to the target operation rate.

[3. An example of the flow of processing executed by the information providing device]
Next, the flow of the extraction process executed by the information providing device 10 will be described using FIG. 5. FIG. 5 is a flowchart illustrating an example of the flow of an extraction process executed by the information providing apparatus according to the embodiment.

For example, the information providing device 10 acquires the user's behavior history (step S101). Then, the information providing device 10 specifies the characteristics of the content to be distributed to the user (step S102). After acquiring the user's behavior history and specifying the content features, a similarity score of the user's behavior history with respect to the content features is calculated (step S103). After calculating the similarity score, the information providing device 10 creates an ID list 50 of users in descending order of similarity score (step ST104).

Furthermore, the information providing device 10 calculates the user's operation rate for the distributed content (step S105). After calculating the user's operation rate, the information providing device 10 extracts users with a high operation rate (step S106). That is, the information providing device 10 extracts high operation rate IDs 55, which are IDs of users with high similarity scores and high operation rates, from among the multiple users included in the ID list 50. After extracting the high operation rate ID 55, content is distributed to the extracted user ID (step S107). Thereby, the information providing apparatus 10 can distribute the content to users who have a high similarity score and a high interaction rate with respect to the content.

[4. Modified example]
An example of the extraction process by the information providing device 10 has been described above. However, embodiments are not limited thereto. Hereinafter, variations of the extraction process executed by the information providing device 10 will be explained.

[4-1. About the operation rate for content]
In the above example, the extraction unit 45 of the information providing device 10 extracts users with a high click rate when extracting users with a high interaction rate with respect to the content. Users with a high operation rate may be extracted. That is, the operation rate calculation unit 44 included in the control unit 40 of the information providing device 10 calculates the operation rate for each of a plurality of different operations performed by the user on the content, and the extraction unit 45 calculates the similarity score. Users with high operation rates calculated by the operation rate calculation unit 44 may be extracted from among the multiple users with high operation rates.

FIG. 6 is an explanatory diagram when extracting IDs of users with high advertisement click rate and advertisement conversion rate. The operation rate calculation unit 44 may calculate an advertisement conversion rate (CVI) in addition to the advertisement click rate (CTR), as shown in FIG. 6, for example, as the operation rate for the content. In this case, the extraction unit 45 sets the lower limit CTR as a threshold for CTR, sets the lower limit CVI as a threshold for CVI, and the similarity score is the threshold set for the similarity score. Among the IDs of a plurality of users whose CTR is equal to or higher than the lower limit similarity score, the ID of a user whose CTR is equal to or higher than the lower limit CVI and whose CVI is the lower limit CVI is extracted as a high operation rate ID55.

Further, the extraction unit 45 changes and adjusts the lower limit CTR so that the user's average CTR approaches the target CTR, and changes and adjusts the lower limit CVI so that the user's average CVI approaches the target CVI. Thereby, it is possible to extract a high operation rate ID55 that can bring the average CTR and average CVI closer to the target CTR and target CVI.

[4-2. Regarding the types of operation rates for content]
In addition, when extracting users with high interaction rates from multiple users with high similarity scores, the interaction rate is determined by the advertisement click rate (CTR) and advertisement conversion rate (CVI). It may be other than that.

FIG. 7 is an explanatory diagram when extracting IDs of users with high index 1 and index 2 of operation rate. In addition to the advertising click rate (CTR) and advertising conversion rate (CVI), the interaction rate for content is assumed to be the interaction rate for various operations. When extracting users with a high rate, it is preferable to set a lower limit value and a target value for each type of operation rate. For example, if multiple operation rates when extracting users with a high operation rate are index 1 and index 2 as shown in FIG. 7, for index 1, the lower limit value of index 1 and the target value is set, and for index 2, the lower limit value and target value of index 2 are set. In this case, the extraction unit 45 determines that among the IDs of a plurality of users whose similarity scores are equal to or higher than the lower limit similarity score set for the similarity score, index 1 is the lower limit value of index 1 and index 2 is the lower limit value of index 1. The user ID that is the lower limit value of index 2 is extracted as high operation rate ID55.

The extraction unit 45 also changes and adjusts the lower limit value of index 1 so that the average value of index 1 approaches the target value of index 1, and adjusts the lower limit value of index 1 so that the average value of index 2 approaches the target value of index 2. Adjust by changing the lower limit value of 2. Thereby, a high operation rate ID55 that can bring the average value of index 1 and the average value of index 2 close to the target value of index 1 and the target value of index 2 can be extracted.

When extracting a user with a high number of operation rates from among multiple users with a high similarity score, the extraction is performed using the similarity score and multiple indexes of operation rates. As shown in the figure, by creating the ID list 50 in the number N+1-dimensional space of a plurality of operation rate indicators, high operation rate IDs 55 can be extracted.

[4-3. Device configuration〕
The information providing device 10 may be realized by a front end server and a back end server. In such a case, the front-end server is provided with the acquisition section 41 and the identification section 42 shown in FIG. 2, and the back-end server is provided with the score calculation section 43, the operation rate calculation section 44, and the extraction section 45. . Furthermore, each of the databases 31 and 32 registered in the storage unit 30 may be held in an external storage server.

[4-4. others〕
Further, among the processes described in the above embodiments, all or part of the processes described as being performed automatically can be performed manually, and conversely, the processes described as being performed manually can be performed manually. All or part of this can also be performed automatically using known methods. In addition, information including the processing procedures, specific names, and various data and parameters shown in the above documents and drawings may be changed arbitrarily, unless otherwise specified. For example, the various information shown in each figure is not limited to the illustrated information.

Further, each component of each device shown in the drawings is functionally conceptual, and does not necessarily need to be physically configured as shown in the drawings. In other words, the specific form of distributing and integrating each device is not limited to what is shown in the diagram, and all or part of the devices can be functionally or physically distributed or integrated in arbitrary units depending on various loads and usage conditions. Can be integrated and configured.

Furthermore, the embodiments described above can be combined as appropriate within a range that does not conflict with the processing contents.

[4-5. program〕
Further, the information providing apparatus 10 according to the embodiment described above is realized by, for example, a computer 1000 having a configuration as shown in FIG. FIG. 8 is a diagram showing an example of the hardware configuration. The computer 1000 is connected to an output device 1010 and an input device 1020, and has an arithmetic device 1030, a primary storage device 1040, a secondary storage device 1050, an output IF (Interface) 1060, an input IF 1070, and a network IF 1080 connected by a bus 1090. has.

The arithmetic unit 1030 operates based on programs stored in the primary storage device 1040 and the secondary storage device 1050, programs read from the input device 1020, and performs various processes. The primary storage device 1040 is a memory device such as a RAM that temporarily stores data used by the arithmetic unit 1030 for various calculations. Further, the secondary storage device 1050 is a storage device in which data used by the arithmetic unit 1030 for various calculations and various databases are registered, and includes ROM (Read Only Memory), HDD (Hard Disk Drive), flash memory, etc. This is realized by

The output IF 1060 is an interface for transmitting information to be output to an output device 1010 that outputs various information such as a monitor or a printer, and is, for example, a USB (Universal Serial Bus), a DVI (Digital Visual Interface), This is realized using a connector compliant with standards such as HDMI (registered trademark) (High Definition Multimedia Interface). Further, the input IF 1070 is an interface for receiving information from various input devices 1020 such as a mouse, a keyboard, and a scanner, and is realized by, for example, a USB or the like.

Note that the input device 1020 is, for example, an optical recording medium such as a CD (Compact Disc), a DVD (Digital Versatile Disc), or a PD (Phase change rewritable disk), a magneto-optical recording medium such as an MO (Magneto-Optical disk), or a tape. It may be a device that reads information from a medium, a magnetic recording medium, a semiconductor memory, or the like. Further, the input device 1020 may be an external storage medium such as a USB memory.

Network IF 1080 receives data from other devices via network N and sends it to computing device 1030, and also sends data generated by computing device 1030 to other devices via network N.

Arithmetic device 1030 controls output device 1010 and input device 1020 via output IF 1060 and input IF 1070. For example, the arithmetic device 1030 loads a program from the input device 1020 or the secondary storage device 1050 onto the primary storage device 1040, and executes the loaded program.

For example, when the computer 1000 functions as the information providing device 10, the arithmetic unit 1030 of the computer 1000 realizes the functions of the control unit 40 by executing a program loaded onto the primary storage device 1040.

[5. effect〕
As described above, the information providing device 10 calculates a similarity score for each user based on the user's action history, which indicates the degree of similarity between the user's action history and the content feature, and distributes the similarity score to the user. An operation rate is calculated for each user, which indicates the rate at which users interacted with the content. Then, the information providing device 10 extracts a user with a high operation rate from among the multiple users with a high similarity score. Thereby, the information providing device 10 can extract users who can enhance the content distribution effect. Therefore, by distributing content to users extracted by the information providing device 10, it is possible to distribute content to users who are unlikely to perform operations on the content even if the similarity score of their behavior history with respect to the features of the content is high. Distribution of content can be suppressed.

In addition, the information providing device 10 calculates operation rates for each of a plurality of different operations performed by the user on the content, and selects one of the plurality of operation rates from among the plurality of users with a high similarity score. Extract users with a high level of comfort. Therefore, even if a plurality of types of operations are assumed to be performed on the content, it is possible to extract users who have a high operation rate for each operation. Thereby, the information providing device 10 can extract users who can enhance the content distribution effect.

Further, the information providing device 10 extracts a plurality of users who can bring the average operation rate, which is the average operation rate of the plurality of users, closer to the target operation rate, which is the target operation rate. Therefore, it is possible to effectively extract users who have a high operation rate with respect to content. Thereby, the information providing device 10 can extract users who can enhance the content distribution effect.

Furthermore, the information providing device 10 extracts users whose operation rate is equal to or higher than a threshold set for the operation rate as users with a high operation rate, and thus appropriately extracts users who have a high operation rate with respect to the content. be able to. Thereby, the information providing device 10 can extract users who can enhance the content distribution effect.

Furthermore, the information providing device 10 brings the average operation rate of the plurality of users to be extracted closer to the target operation rate by changing the operation rate threshold. Therefore, it is possible to effectively extract users who have a high operation rate with respect to content. Thereby, the information providing device 10 can extract users who can enhance the content distribution effect.

Furthermore, the information providing device 10 extracts users with a high operation rate from among users whose similarity score is equal to or greater than a threshold set for the similarity score. Therefore, it is possible to appropriately extract users who have a high similarity score with respect to the features of the content and a high operation rate with respect to the content. Thereby, the information providing device 10 can extract users who can enhance the content distribution effect.

Furthermore, the information providing device 10 brings the average operation rate of the plurality of users to be extracted closer to the target operation rate by changing the threshold of the similarity score. Therefore, it is possible to effectively extract users who have a high similarity score with respect to the features of the content and a high operation rate with respect to the content. Thereby, the information providing device 10 can extract users who can enhance the content distribution effect.

Further, the information providing device 10 sets the value calculated based on the multiplication of the similarity score and the operation rate to a lower limit set with respect to the value calculated based on the multiplication of the similarity score and the operation rate. Users with a high operation rate are extracted from the above users. Therefore, it is possible to appropriately extract users who have a high similarity score with respect to the features of the content and a high operation rate with respect to the content. Thereby, the information providing device 10 can extract users who can enhance the content distribution effect.

Some of the embodiments of the present application have been described above in detail based on the drawings, but these are merely examples, and various modifications and variations may be made based on the knowledge of those skilled in the art, including the embodiments described in the disclosure section of the invention. It is possible to carry out the invention in other forms with modifications.

Furthermore, the above-mentioned "section, module, unit" can be read as "means", "circuit", etc. For example, the providing unit can be read as providing means or providing circuit.

10 information providing device 20 communication unit 30 storage unit 31 user database 32 content database 40 control unit 41 acquisition unit 42 identification unit 43 score calculation unit 44 operation rate calculation unit 45 extraction unit 50 ID list 55 high operation rate ID
56 curved section 100 log server 200 distribution server 300 terminal device

Claims (10)

an acquisition unit that acquires the user's action history;
a specifying unit that specifies the characteristics of the content to be distributed to the user;
a score calculation unit that calculates a similarity score for each of the users based on the action history, which indicates the degree of similarity of the action history with respect to the feature of the content;
an operation rate calculation unit that calculates, for each user, an operation rate indicating the rate at which the user performed operations on the content distributed to the user;
an extraction unit that extracts the user with the high operation rate from among the multiple users with the high similarity score;
An extraction device characterized by having: The operation rate calculation unit calculates the operation rate for each of a plurality of different operations performed by the user on the content,
2. The extraction unit extracts the users with high all of the plurality of operation rates calculated by the operation rate calculation unit from among the plurality of users with high similarity scores. Extraction device. 2. The extraction unit extracts a plurality of users who can bring an average operation rate, which is the average operation rate of the plurality of users, closer to a target operation rate, which is the operation rate as a target. Or the extraction device according to 2. The extraction device according to claim 3, wherein the extraction unit extracts the user whose operation rate is equal to or higher than a threshold set for the operation rate as the user who has the high operation rate. The extraction device according to claim 4, wherein the extraction unit brings the average operation rate of the plurality of users to be extracted closer to the target operation rate by changing the threshold value of the operation rate. The extraction device according to claim 3, wherein the extraction unit extracts the users with the high operation rate from among the users whose similarity scores are equal to or higher than a threshold set for the similarity scores. The extraction device according to claim 6, wherein the extraction unit brings the average operation rate of the plurality of users to be extracted closer to the target operation rate by changing the threshold value of the similarity score. The extraction unit is configured such that a value calculated based on the multiplication of the similarity score and the operation rate is a lower limit set with respect to a value calculated based on the multiplication of the similarity score and the operation rate. The extraction device according to claim 3, wherein the user whose operation rate is high is extracted from among the users whose operation rate is higher than a value. An extraction method performed by an extraction device, comprising:
an acquisition step of acquiring the user's behavior history;
a specifying step of specifying the characteristics of the content to be distributed to the user;
a score calculation step of calculating a similarity score for each user based on the action history, which indicates the degree of similarity of the action history to the feature of the content;
an operation rate calculation step of calculating for each user an operation rate indicating the rate at which the user performed an operation on the content distributed to the user;
an extraction step of extracting the user with the high operation rate from among the multiple users with the high similarity score;
An extraction method characterized by comprising: an acquisition procedure for acquiring the user's behavior history;
a specific procedure for specifying the characteristics of the content to be distributed to the user;
a score calculation procedure for calculating a similarity score for each user based on the action history, which indicates the degree of similarity of the action history to the feature of the content;
an operation rate calculation procedure for calculating, for each user, an operation rate indicating the rate at which the user performed operations on the content distributed to the user;
an extraction procedure for extracting the user with the high operation rate from among the multiple users with the high similarity score;
An extraction program that allows a computer to execute

Images