JP2023089233A - Information processing apparatus and program - Google Patents

Abstract

To provide an information processing apparatus and program for properly creating a persona in order to improve the accuracy of refining target users of advertisement distribution and to implement effective advertisement distribution.SOLUTION: In an overall system, an information processing apparatus 1 connected via an Internet line 8 to an SNS server 2 that provides an SNS service includes: means for requesting information on SNS users who are provided with the SNS service from the SNS server 2 and downloading the information; and means for extracting, from among the SNS users, based on the information, an SNS user who is interested in a specific SNS user or a post thereof and who has a persona which is the same as or similar to a typical persona related to an SNS user who has posted the same commodity as a commodity posted by the specific SNS user.SELECTED DRAWING: Figure 2

Description

TECHNICAL FIELD Embodiments of the present invention relate to an information processing apparatus and a program.

Recently, it has become difficult to have a sales form (physical store sales) in which a salesperson sells merchandise while serving customers in a store. The ratio of actual store sales to total sales is decreasing, and the ratio of sales forms (EC sales) using electronic commerce (EC: Electronic Commerce) via the Internet is increasing.

Furthermore, SNS (Social Networking Service) can be used to attract customers to EC sites, that is, specific SNS users (image contributors) such as sales staff at physical stores can post product photos (images) on Instagram (registered trademark), for example. ) and guide SNS users to the EC site from the posting, thereby leading to product sales.

This form of sales utilizing SNS will shift from the so-called "waiting sales" where SNS users wait for an opportunity to visit the poster's posting page, and distribute advertisements individually to SNS users to guide them to the posting page. We are shifting to active sales activities that There are various types of advertisements such as display advertisements and mail advertisements. In both cases, advertisements are distributed by narrowing down the target users of the advertisement distribution, but the success rate such as click-through rate and conversion rate is not very high.

One of the reasons for this is considered to be that the accuracy of narrowing down target users for advertisement distribution is not very high.

The purpose is to improve the accuracy of narrowing down the target users of advertisement distribution and to create personas that include interests as well as profiles in order to realize effective advertisement distribution.

This embodiment is an information processing apparatus connected via an Internet line to an SNS server that provides SNS services to a plurality of SNS users, and includes a storage unit that stores a program, and an information processing apparatus that executes the program. a processor and a communication unit that communicates with the SNS server via the Internet line, and the processor executes the program to communicate with the SNS server via the communication unit to each of the SNS users profile of each of the SNS users, posted images posted by each of the SNS users, hashtags attached to the posted images by each of the SNS users, means for requesting and downloading comments posted by each of the SNS users, and the hashtags and the comments. means for extracting nouns by performing language analysis processing on the SNS user, and based on the extracted nouns, identifying the classification of the target of interest of each of the SNS users; and means for creating persona data for each SNS user.

FIG. 1 is a diagram showing a screen configuration example of an SNS service. FIG. 2 is a configuration diagram of an entire system including an information processing apparatus according to this embodiment. FIG. 3 is a configuration diagram of the information processing apparatus of FIG. FIG. 4 is a diagram showing an example of a persona. FIG. 5 is a diagram illustrating an example taxonomy of interests identified by the processor of FIG. FIG. 6 is a flow chart showing a persona creation processing procedure implemented by the processor of FIG. FIG. 7 is a conceptual diagram of the persona creation processing of FIG. FIG. 8 is a flowchart showing a typical persona creation processing procedure of an SNS user who has a strong relationship with user A, which is realized by the processor of FIG. FIG. 9 is a conceptual diagram of the typical persona creation process of FIG. FIG. 10 is a flow chart showing the procedure of purchase candidate extraction processing (human axis) implemented by the processor of FIG. FIG. 11 is a conceptual diagram of purchase candidate extraction in FIG. FIG. 12 is a flow chart showing the procedure of purchase candidate extraction processing (product axis 1) implemented by the processor of FIG. FIG. 13 is a conceptual diagram of purchase candidate extraction in FIG. FIG. 14 is a flow chart showing the procedure of purchase candidate extraction processing (product axis 2) implemented by the processor of FIG. FIG. 15 is a conceptual diagram of purchase candidate extraction in FIG.

Embodiments of the present invention will be described below with reference to the drawings.
This embodiment uses "Facebook (registered trademark)", "Twitter (registered trademark)", " It relates to the process of extracting product purchase candidates by utilizing image (still image (photo), video) posting and viewing services such as Instagram (registered trademark). As shown in FIG. 1, an example of the screen of the image posting and browsing service, an SNS user who receives the SNS service registers a profile and uploads photos and videos (hereinafter referred to as images) related to hobbies, sales products, etc. to themselves. A function is provided to post along with the comments of On the other hand, SNS users who browse posted images and the like are provided with a function that allows them to react to the users who posted the images (image posters) and the posts, and to execute some actions. A function that allows you to grasp the update status of the image poster by following the person, a function to send comments on the post, a function to send a sympathetic intention such as "like" to the post, a function to your own followers, etc. Various functions such as functions such as retweeting and sharing to recommend and spread posts etc., and functions to save (download) posted images are provided.

In addition, by executing actions such as follow function, comment posting function, function to express sympathy such as "Like", diffusion function such as retweet and share, save function, etc., react to the person who posted the image or their post However, an SNS user who performs some action is considered to be an SNS user who is attracted to the target image poster or his/her post and is showing a strong interest. Here, the SNS user who performs at least one of the follow function, comment posting function, sympathetic intention transmission function such as "Like", retweet and share function, and storage function is the image poster or The user is said to be an SNS user who shows an interest in the post.

As shown in FIG. 2, an SNS server 2 that provides SNS services is connected to an information processing apparatus 1 according to the present embodiment via a public communication line network (Internet line) 8 . Terminals (user terminals) 3 - 7 such as PCs, smartphones, PHSs, PDAs, tablet terminals, etc. owned by SNS users such as image posters and viewers are connected to the SNS server 2 via the Internet line 8 .

For example, an SNS user as an image poster can post images, comments, etc. to the SNS server 2 via the user terminal 3 . On the other hand, SNS users as viewers can comment on posts, sympathize with posts such as “likes”, dislike posts such as “I hate” posts, spread posts such as retweets and shares, save posted images (download ), followers of the poster, etc., can transmit various reactions via the user terminal 4-7.

As shown in FIG. 3, the information processing apparatus 1 is connected to a processor 11 via a data/control bus 10 with a RAM 12, a ROM 13, a storage section 14, an input controller 15, a video controller 16, and a communication section 17. FIG. An input device 18 such as a keyboard and a mouse is connected to the input controller 15 . A display 19 such as an LCD (Liquid Crystal Display) is connected to the video controller 16 .

The processor 11 is composed of a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit). The processor 11 executes various programs loaded from the storage unit 14 or the ROM 13 to the RAM 12 to execute persona creation processing, typical persona creation processing, and three types of purchase candidate extraction processing, which will be described later. The RAM 12 functions as a main memory, work area, etc. of the processor 11 . The ROM 13 or the storage unit 14 stores a BIOS (Basic Input Output System) executed by the processor 11, an operating system program (OS), a persona creation program according to the present embodiment, a typical persona creation program, and three types of purchase candidates. An extraction program, other programs for realizing various functions such as preprocessing, and various data required for these processes are stored.

Next, persona creation processing will be described. As illustrated in FIGS. 4(a) to 4(c), a persona is defined as a persona of an SNS user. It is set from each item in the trend that represents the object shown. Further, as exemplified in FIG. 5, the items of interest, which are the items most related to product purchasing behavior, are more specifically composed of major categories such as fashion and automobiles, lower middle categories, and small categories.

Various means related to persona creation processing are realized by the processor 11 executing a persona creation program loaded from the storage unit 14 or the ROM 13 to the RAM 12 .

As shown in FIGS. 6 and 7, first, in step S11, a profile registered by each SNS user, a posted image posted by each SNS user, and an image attached to the posted image are sent to the SNS server 2 via the communication unit 17. Requests and downloads the hashtags and comments posted by each SNS user on their own posts. The profile data separated by the separating means 302 is passed to the persona creating means 307 , the posted image data is passed to the object extracting means 303 , and the hash tags and comments are passed to the language analyzing means 305 .

In step S12, the language analysis means 305 performs language analysis processing (text mining) on the downloaded hashtags and comments to extract nouns, and the interest target identification means 306 performs SNS based on the extracted nouns. Identify broad categories of user interests. As the specific processing, a correspondence table between a large number of nouns and a plurality of major categories is constructed in advance, nouns are individually referred to the correspondence table, and the major category of interest that indicates the maximum frequency is selected as the interest of the SNS user. It may be specified as a large classification of the target, or a plurality of nouns may be input by a trained neural network, and a large classification of the target of interest may be output.

In step S13, the object extracting means 303 extracts a partial image including an object corresponding to the major category of interest, for example, a jacket corresponding to fashion, from the posted image using the specified major category of interest. A convolutional neural network is typically applied to this extraction process. Since the range of image analysis processing, which will be described later, can be limited to partial images containing objects corresponding to the major classification of the target of interest, improvement in image analysis accuracy can be expected.

The image analysis means 304 is divided into a plurality of identification means 304-1, 304-2, . The plurality of discriminating means 304-1, 304-2, . . . 304-n are typically constructed by a convolutional neural network system. As is well known, a convolutional neural network extracts a feature amount from an image by convolving a two-dimensional image with a filter. A convolutional neural network consists of a multilayer network that repeats convolution and pooling. Coefficients (weights) of filters effective for identification that constitute convolutional layers in a convolutional neural network are learned using a large amount of data such as a large amount of training images. The coefficients (weights) are obtained by learning to acquire invariance to various deformations by repeating convolution with a filter and pooling of responses in a certain region using a large amount of data. Note that the identification performance of the convolutional neural network depends on the filters that constitute the convolutional layers. A filter is prepared for each major classification of the object of interest.

In step S14, one of the plurality of identification means 304-1, 304-2, . Estimate the lower middle and minor classifications of the target. If the object of interest (major classification) is "fashion", a classifier constructed to match the major classification "fashion" is selected, and the selected discriminator analyzes the partial image and classifies it as a middle class. "Cool" and "high brand" as a minor classification are output (see FIG. 4(a)).

Note that when there are a plurality of posted images posted by the SNS user, steps S12, S13, and S14 are repeated for each posted image. Since the major, intermediate, and minor categories of interest are specified for each posted image, there may be multiple major, intermediate, and minor categories. Classifications and minor classifications are identified as major, middle and minor classifications of representative interests of the SNS user.

In step S15, the trend analysis means 308 analyzes the SNS based on the language analysis results of the hashtags and comments attached to the latest or a predetermined number of recent posted images among the plurality of posted images posted by the SNS user. Identify trends that represent what your users are currently interested in. For example, among the multiple posted images posted by the SNS user, the hashtag attached to the latest one or a predetermined number of posted images, and from the own comment, multiple nouns are extracted by language analysis processing, and the appearance of each noun Aggregate the frequencies and identify the nouns that show the highest frequency as trends. For example, when the noun "Korean fashion" has the highest frequency, "Korean fashion" is identified as the user's trend (see FIG. 4A).

In step S16, the persona creation means 307 creates persona data by associating age, gender, place of residence, interests (major category, middle category, minor category), and trends from specific items in the profile with the account of the SNS user. do. The persona data is stored in the storage unit 14 in step S17.

Steps S11-S17 are repeated through step S18 to create and store persona data for all SNS users who receive the SNS service or for SNS users included in a preset specific group.

Next, referring to FIGS. 8 and 9, processing for creating a typical persona of an SNS user (D) who has a strong interest in a specific user A and who also has a strong interest in products posted by user A. will be explained. Here, it is assumed that user A is a salesclerk working at an apparel shop, an SNS user, and an image poster who posts images of various products sold at the apparel shop.

In step S21, the SNS server 2 is requested via the communication unit 17 for the account of the user A or the user B who shows a strong interest in the post, and is downloaded. Specifically, User A or User B who shows a strong interest in the post of User A can be a follower of User A, a user who expresses an intention to empathize with User A's post by retweeting or sharing it. A user who diffused A's post, a user who saved (downloaded) an image posted by user A, a user who posted a comment on user A's post, and the like.

In step S<b>22 , a request is made to the SNS server 2 via the communication unit 17 for the hash tag attached to the posted image of the user A, and the hash tag is downloaded.

In step S23, the name of the product (product name) posted by User A is specified based on the language analysis processing result of the hashtag attached to User A's posted image. Note that the name of the product (product name) posted by the user A may be input by the operator via the input device 18 .

In step S24, the SNS server 2 is requested via the communication unit 17 for the account of the user C who posted the image with a hashtag containing the product name (product name) posted by the user A, and the image is downloaded. do. Since user C has posted an image of the same product as user A's posted product, it is presumed that user C has a strong interest in user A's posted product.

In step S25, a plurality of users D in which the user B who shows a strong interest in the user A or its contribution and the user C who shows a strong interest in the product posted by the user A overlap are extracted. It can be said that user D is a user who shows a strong interest in both user A or its contribution and user A's contribution products.

In step S26, the persona data of user D is read out from the storage unit 14, and in step S27, the persona of user D is integrated and a typical persona (PA) of user D is created. The items that make up a persona are aggregated by age group, gender, place of residence, subject of interest (major category), subject of interest (middle category), subject of interest (minor category), and trend. The age, sex, place of residence, target of interest (major category), target of interest (middle category), target of interest (minor category), and specific contents of trends are specified. A typical persona is an element that characterizes the personal image shared by a plurality of users D who show a strong interest in user A or its posts and who also show a strong interest in products posted by user A. It can be said that it is a group.

In step S28, the typical persona data is stored in the storage unit 14 in association with user A's account.

Next, a purchase candidate extraction process will be described. As for the circumstances leading up to purchasing products using SNS services, User A began to show a strong interest in User A and his posts. It is assumed that the user A will show a strong interest, move to the EC site via the posting by the user A, and purchase the product sold on the EC site.

In the present embodiment, users who are likely to show a strong interest in user A or his posts in the future are extracted as purchase candidates, and purchase candidates are extracted mainly based on the persona of a specific user A. Two types of methods are provided: a method (referred to as a human-axis method) and a method of extracting purchase candidates starting from products posted by user A (referred to as a product-axis method). Furthermore, two types are provided as the latter product axis method. In this embodiment, these human-axis methods and two types of product-axis methods may be provided and selectively applied, or all of the human-axis method and two types of product-axis methods or any two methods. You may narrow down by logical product of the purchase candidate extracted by.

10 and 11 show the processing procedure of the human axis method.
In step S31, the persona data of the SNS user created in advance, that is, the persona data of the specific user A and the other users are read out from the storage unit 14. User E, who has the same or similar persona as user A, is extracted. User E has the same age, gender, place of residence, target of interest (major category), target of interest (middle category), target of interest (minor category), and all items of trend as user A. Alternatively, user E differs from user A in some items of age, gender, place of residence, target of interest (major category), target of interest (middle category), target of interest (minor category), and trend. When the items are the same. Here, as the different items, items other than the items considered to be important for showing a strong interest in a person are set in advance, and for example, a place of residence and an object of interest (minor classification) are set. In other words, the place of residence and the target of interest (small category) are different from those of user A, but the other items such as age, gender, target of interest (major category), target of interest (middle category), and trend are the same as those of user A. is extracted.

In step S33, the SNS server 2 is requested via the communication unit 17 for the account of the user E or the user F who shows a strong interest in the post, and is downloaded. That is, it is assumed that the user F who shows a strong interest in the user E whose personal image is similar to that of the specific user A is likely to show a strong interest in the specific user A as well. As described above, User E or User F who shows a strong interest in the post of User E includes followers of User E, users who express their sympathy for User E's post such as "Like", retweet, share, etc. , a user who saved (downloaded) an image posted by user E, a user who posted a comment on user E's post, and the like.

In step S34, user G is extracted from user F by excluding user D, who has already shown a strong interest in user A or his contribution, and who has also shown a strong interest in user A's posted products. It is assumed that user G has a strong interest in user E, who is similar in character to user A, but has never had contact with user A until now.

In step S35, data of a typical persona (PA) of user A created in advance from the storage unit 14 or a typical persona (PA) of user D who shows a strong interest in the post by the user A and also shows a strong interest in the product posted by user A is retrieved. From the user G, extract the user H who has the same or similar persona as the typical persona (PA). Here, “similar” means, as described above, age, gender, place of residence, target of interest (major category), target of interest (middle category), target of interest (minor category), and some of the items of trend Different items refer to cases where other items are the same, and as different items, items other than items considered to be important in showing a strong interest in a person are set in advance, such as a place of residence and an object of interest (small classification). is set.

User H is assumed to be a user who is highly likely to show a strong interest in a particular user A and purchase products posted by user A in the future. In other words, it is very likely that user H will become a follower of a specific user A in the future, or will respond to user A's post by sending a "like" or other sympathetic intention. It can be said that there is a high possibility that the user will move to the EC site via the posting and finally purchase the product.

12 and 13 show the processing procedure of the first product axis method.
First, in step S41, a request is made to the SNS server 2 via the communication unit 17 for the hash tag attached to the posted image posted by the specific user A, and the hash tag is downloaded. In step S42, the names (product names) of products a, b, and c posted by user A are specified based on the language analysis processing result of the hashtags. Note that the name of the product (product name) posted by the user A may be input by the operator via the input device 18 . In step S<b>43 , names of similar products d and e similar to products a, b and c posted by user A are input by the operator via the input device 18 . As similar products d and e, the names of competing products of products a, b, and c posted by user A are typically input.

In step S44, a request is made to the SNS server 2 via the communication unit 17 for the account of the user J who posted the image with the hashtag including the names of the products a, b, and c posted by the user A. ,to download. Similarly, in step S45, user K posted an image to the SNS server 2 via the communication unit 17 with a hash tag including the names of similar products d and e similar to the product posted by user A. Request an account to download. Since users J and K have posted products a, b, c or similar products d, e posted by user A, they have already purchased products a, b, c, d, and e. Even if it does not reach to , it is assumed that the user A has a strong interest, and that the products handled by the apparel shop where the user A works relatively match the user's taste.

In step S46, user A or user D, who has already shown a strong interest in user A's contribution, is excluded from users J and K, and although user A has shown a strong interest in the product posted by user A or a similar product, user J It is assumed to be a user who had no contact with A.

In step S47, data of a typical persona (PA) of user A created in advance or of user D who has a strong interest in the post and a strong interest in the posted product or similar product is read out from the storage unit 14. , users L, extract users M who have the same or similar persona as the typical persona (PA). Here, "similar" is as described above.

User L is very likely to show a strong interest in products posted by specific user A in the future, and shows interest in products posted by specific user A. It is assumed that the purchase candidate is highly likely to move to the EC site via the website and finally purchase the product.

14 and 15 show the processing procedure of the second product axis method.
First, in step S51, a request is made to the SNS server 2 via the communication unit 17 for the hash tag attached to the posted image posted by the specific user A, and the hash tag is downloaded. In step S52, the names (product names) of products a, b, and c posted by user A are specified based on the language analysis processing result of the hashtags. Of course, the name of the product (product name) posted by the user A may be input by the operator via the input device 18 .

In step S53, a request is made to the SNS server 2 via the communication unit 17 for the account of the user J who posted the image with the hashtag containing the names of the products a, b, and c posted by the user A. ,to download.

In step S54, user N is extracted from user J by excluding user A who downloaded in step S21 or user B who shows a strong interest in the post. It is assumed that user N has shown a strong interest in user A's posted products, but has never had contact with user A until now.

In step S55, data of a typical persona (PA) of user A or user D who has shown a strong interest in the post created in advance is read from the storage unit 14, and the typical persona (PA) is read from user N. Extract a user P who has the same or similar persona as . Here, "similar" is as described above.

User P is very likely to show a strong interest in products posted by specific user A in the future, and shows interest in products posted by specific user A. It is assumed that the purchase candidate is highly likely to move to the EC site via the website and finally purchase the product.

According to this embodiment, there is a very high possibility that a specific user A will show a strong interest in the future, and there is a possibility that user A will move to an EC site via a post by user A and finally purchase a product. are likely to show a strong interest in the product posted by User A, and move to the e-commerce site via User A's post and finally purchase the product. It is possible to extract purchase candidates who are likely to purchase. Therefore, it is possible to realize effective advertisement distribution to these purchase candidates.

While several embodiments of the invention have been described, these embodiments have been presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and modifications can be made without departing from the scope of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, as well as the scope of the invention described in the claims and equivalents thereof.

1... Information processing device, 2... SNS server, 3-7... User terminal.

Claims (10)

An information processing device connected via an Internet line to an SNS server that provides SNS services to a plurality of SNS users,
a storage unit that stores a program;
a processor that executes the program;
a communication unit that communicates with the SNS server via the Internet line,
By the processor executing the program,
a profile of each of the SNS users, a posted image posted by each of the SNS users, a hashtag attached to the posted image by each of the SNS users, and a post by each of the SNS users to the SNS server via the communication unit; a means for requesting and downloading comments made;
A means for performing language analysis processing on the hashtags and the comments to extract nouns, and identifying a classification of interest targets of each of the SNS users based on the extracted nouns;
An information processing apparatus that realizes means for creating persona data of each of the SNS users from the profile and the classification of the target of interest. By the processor executing the program,
further realizing image analysis means for performing image analysis processing suitable for the classification of the target of interest on the posted image to identify a lower classification of the target of interest;
2. The information processing apparatus according to claim 1, wherein said persona data is created from said profile, said interest classification, and a sub-classification of said interest. 2. The information processing apparatus according to claim 1, wherein said image analysis means is configured by a convolutional neural network. By the processor executing the program,
Further realizing a means for identifying trends of the SNS users based on a hashtag attached to the latest one or a predetermined number of posted images among the posted images and language analysis processing results of the comments,
The information processing apparatus according to claim 1, wherein the persona data is created from trends of the SNS user in addition to the profile and the classification of the subject of interest. An information processing device connected via an Internet line to an SNS server that provides SNS services to a plurality of SNS users,
a storage unit that stores a program and persona data of each of the SNS users;
a processor that executes the program;
a communication unit that communicates with the SNS server via the Internet line,
By the processor executing the program,
means for requesting and downloading an account of a first SNS user showing interest in a specific SNS user from the SNS server via the communication unit;
means for requesting and downloading an account of a second SNS user who is interested in the product posted by the specific SNS user to the SNS server via the communication unit;
means for extracting a third SNS user who overlaps the first SNS user and the second SNS user;
a means for creating typical persona data relating to SNS users who are interested in the specific SNS user and products posted by the specific SNS user, based on the persona data of each of the third SNS users. , information processing equipment. The first SNS user includes a follower of the specific SNS user, an SNS user who expressed an intention to empathize with the post of the specific SNS user, an SNS user who spread the post of the specific SNS user, and the specific SNS. 6. The information processing apparatus according to claim 5, wherein the user is at least one of an SNS user who saved an image posted by the user and an SNS user who commented on the post by the specific SNS user. 6. The information processing apparatus according to claim 5, wherein the second SNS user is an SNS user who posted an image including a hash tag containing a name of a product posted by the specific SNS user. By the processor executing the program,
means for executing a language analysis process on a hashtag attached to an image posted by the specific SNS user;
6. The information processing apparatus according to claim 5, further comprising means for identifying the name of the product posted by said specific SNS user based on the result of said language analysis processing. 6. The information processing apparatus according to claim 5, wherein said means for creating said typical persona data aggregates the personas of each of said third SNS users for each item, and creates said typical persona from content indicating maximum frequency. . A program for causing a computer to function as the information processing apparatus according to any one of claims 1 to 9.

Images