JP7503804B2 - Information processing device and program - Google Patents

Description

An embodiment of the present invention relates to an information processing device and a program for predicting the sales ability of a salesperson selling over the Internet.

Recently, it has become difficult to continue selling products in a store while sales staff are serving customers (physical store sales). The proportion of physical store sales in total sales is decreasing, and the proportion of sales using electronic commerce (EC sales) via the Internet is increasing.

Furthermore, it is becoming common to use SNS (social networking services) to attract customers to EC sites. In other words, sales staff (contributors) at physical stores post product images, etc. on Instagram (registered trademark), for example, and then direct users to the EC site from those posts.

However, unlike retail stores, predicting customer attraction and sales power using SNS is extremely difficult as there are currently no established methods, and the accuracy is not high. In addition, there is confusion as to what is lacking and what activities should be done to use SNS to attract more customers and increase sales, and it is difficult to educate not only posters but also companies on how to use SNS.

The aim is to make it possible to evaluate the various capabilities of those who post images on social media.

The information processing device according to the present embodiment is connected to an SNS server that provides an SNS via an Internet line. The information processing device includes a means for requesting and receiving from the SNS server a profile of each follower who follows a poster who posted a plurality of posted images, and a profile of each of a plurality of engagement users who performed a plurality of types of engagement (rating, commenting, saving the posted image) for each of the posted images, and a means for calculating a consistency score representing a consistency of the followers and the engagement users with a target demographic that is preset for a specific product in the posted images, based on the profiles of the followers and the profiles of the engagement users.

FIG. 1 is a diagram showing an example of an SNS screen. FIG. 2 is a configuration diagram of an entire system including a sales force prediction processing device (information processing device) according to this embodiment. FIG. 3 is a block diagram of the sales force forecast processing device of FIG. FIG. 4 is a diagram showing various data stored in the storage device of FIG. FIG. 5 is a diagram showing the input and output of data to and from the neural network realized by the processor of FIG. FIG. 6 is a flowchart showing the operation procedure of information processing in this embodiment. FIG. 7 is a diagram showing the procedure of the estimation process of the product number, the category of the posted image, and the preference tendency in FIG.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. This embodiment relates to a social networking service (hereinafter referred to as "SNS"), which is a community-based service that promotes and supports connections between people, an EC (electronic commerce) service that develops a service for selling products and services on a website via a network such as the Internet, and an Internet service where images (including still images and videos) are posted, such as a video posting and viewing service such as YouTube (registered trademark). Here, the description will be given using as examples SNSs such as "Facebook (registered trademark)", "Twitter (registered trademark)", and "Instagram (registered trademark)". In addition, the description will be given using as an example a case where clothes are sold in a store and via the Web.

These SNSs are highly community-oriented internet services that allow users to post images related to their profile or hobbies along with their own comments, and users who view the images can post comments, give positive impressions such as "like" or "no", negative or sympathetic evaluations, and save the posted images (engagement). Users can also follow posters. Usually, posters can view the number of engagements and the accounts of engaged users and followers from their profile screens. First, a brief explanation of the SNS screen configuration is given. FIG. 1 shows an example of a screen displayed on an Instagram user's terminal. As shown in FIG. 1(a), the screen displays the number of posted images posted by the poster, the number of followers following the poster, the number of users following the poster (following number), the posted images, comments added by the poster to the posted images, positive impressions such as "like" or "no", negative or sympathetic evaluation buttons, a comment button for viewers to post comments on the posted images, an image save button for viewers to save the posted images, and comments posted by viewers on the posted images.

As shown in FIG. 2, a sales management device 2 for managing sales of multiple products through store sales and Web sales, a sales Web server 3 for selling multiple products via a Web page, and an SNS server 4 for providing an SNS are connected to the sales force prediction device 1 according to this embodiment via a public communication line network (Internet line) 9. Terminals (user terminals) 5-8 such as PCs, smartphones, PHS, PDAs, and tablet terminals owned by SNS users such as posters and viewers are also connected to the SNS server 4 via the Internet line 9. For example, posters can post images and comments to the SNS server 4 via the user terminal 5. Engagement users can send comments, ratings such as "like" and "no", and engagement such as saving images on posts, and can also follow posters. Note that the site of the product sales page built on the SNS by the SNS account is substantially the same as the product sales Web site provided by the sales Web server 3.

As shown in FIG. 3, the sales force prediction device 1 is realized by an information processing device, and a RAM 12, a ROM 13, a storage device 14, an input controller 15, a video controller 16, and a communication device 17 are connected to a processor 11 via a system bus 20. An input device 18 is connected to the input controller 15. A display 19 is connected to the video controller 16.

The processor 201 is composed of, for example, a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit). The processor 11 executes a program loaded from the storage device 14 and ROM 13 to the RAM 12 to execute sales force prediction processing. The RAM 12 functions as the main memory, work area, etc. of the processor 11. The ROM 13 or the storage device 14 stores the BIOS (Basic Input Output System) executed by the processor 201, the operating system program (OS), the sales force prediction processing program according to this embodiment, other programs for realizing various functions such as preprocessing, and various data required for these processes.

The input controller 15 controls input from input devices 18, such as a keyboard (KB) and pointing devices such as a mouse or touch panel. The video controller 16, under the control of the processor 11, controls the display of a display 19, such as an LCD (Liquid Crystal Display).

In addition to the sales forecast processing program and pre-processing program, the storage device 14 pre-stores various data required for the sales forecast processing, pre-processing, etc. As shown in FIG. 4(a), data related to product attributes is associated with a product number for identifying the product, and attributes such as the basic color tone of the product, its type (product category) such as whether the product is a top or outerwear, product classification indicating the sales category that distinguishes whether the product is a new product, an older product, or a sale item, and basic design classification. As shown in FIG. 4(b), data related to the product target demographic is associated with the product number, and the gender, age group, etc. targeted by the product.

As shown in FIG. 4(c), data on the posted image category score is set for each category, depending on whether the posted image is an image of the product alone or an image that includes the product in the composition. In more detail, images that include the product are classified into coordinated images of the product combined with other products (clothing), images of models wearing the product in a relaxed state in a landscape, and images of models wearing the product in a workplace. It is assumed that product sales are higher when the posted image is an image of the product in the composition than when the posted image is an image of the product alone, and furthermore, even when the image includes the product, it is assumed that sales are higher when the posted image is an image of a landscape or a workplace than when it is an image of a coordinated outfit. The posted image category score should be set higher in accordance with the expected increase in sales, and the posted image category scores are set individually in advance according to these sales assumptions.

As shown in FIG. 4(d), a score is set for each product individually for the product number as data related to the product-specific score. The product-specific score is a comprehensive, predetermined score that reflects the inherent sales potential of the product, based on the brand power, design power, material quality, etc. In other words, sales are predicted for each product, and the planned sales volume is often set according to the prediction, and a score is preset according to the planned sales volume. The product-specific score should be set high in accordance with the expected increase in sales, and the product-specific score is set individually in advance according to the inherent sales potential and expected sales of the product. It is preferable that the product-specific score is updated appropriately according to the actual sales volume.

In addition, in the prediction process, the period between the time of prediction process and a point a predetermined period of time before that (such as one week or one month) is considered to be the "prediction process target period," and images posted within this prediction process target period are treated as the targets of prediction process.

The sales force forecasting program includes a processor 11 for executing the program,
a means for requesting and receiving from an SNS server 4 a plurality of posted images posted by a poster as a seller of a product within a predicted processing period, a profile of each follower who follows the poster, a plurality of types of engagement numbers for each posted image (number of ratings, number of comments, number of saved posted images), a profile of each of a plurality of engagement users who have performed a plurality of types of engagements for each posted image (ratings, comments, saved posted images), and comments and replies to each posted image;
A means for requesting and receiving sales results of products by contributors from a sales management device (2);
The sales web server 3 is caused to function as a means for requesting and receiving the number of visitors who have flowed into the sales web page via the SNS.

Furthermore, the sales force forecasting program causes the processor 11 to
A means for calculating a reaction score representing the degree of reaction of users to each posted image based on the number of ratings such as the number of likes and dislikes, the number of comments, and the number of saves for the posted image posted within the prediction processing period, and the product-specific score (see FIG. 5( a) );
a means for calculating a consistency score representing the degree of consistency between the product of the target image and the followers and engagement users based on the consistency of the followers and engagement users with a target demographic preset for the product of the posted image within the prediction processing target period and the consistency of the preference trends of the followers and engagement users with the attributes of the product of the target image based on the profiles of the followers and the profiles of the engagement users at the time of the prediction processing (see FIG. 5(b));
a means for calculating a customer service score representing the unique customer service ability of a poster based on the number of all posted images posted by the poster within the prediction processing period, the number of visitors from the pages of those posted images to the sales web page, the number of followers and followings at the time of the prediction processing, the number of comments made on those posted images, the number of replies to those comments, the reply rate to those comments, and the category scores of all posted images posted by the poster within the prediction processing period (see FIG. 5(c));
A means for calculating a sales ability score representing the ability of a poster to sell products through the SNS based on the product score, the consistency score, and the customer service score (see FIG. 5(d) );
It functions as a means (see FIG. 5(e)) for calculating sales forecasts (weekly/monthly/yearly sales forecast amounts, etc.) based on the sales power score, sales performance, and the season at the time of the forecast processing.

These processing means are realized by a neural network in the processor 11. The neural network is machine-trained by periodically inputting the number of engagements, consistency, number of visitors, number of followers, number of followings, number of comments, number of replies to comments, reply rate to comments, category scores of posted images, and the like, together with weekly or monthly sales results, as training data, once a week or once a month or at any frequency.

The program also causes the processor 11 to
A means for determining whether a posted image belongs to a product-only category or a category including a product in its composition by applying a convolutional neural network to each of all posted images posted by a user within a prediction processing target period;
A means for identifying product numbers of products shown in each of all posted images posted by a user within a prediction processing period by applying a convolutional neural network to each of the posted images;
It functions as a means of estimating the taste trends of followers and engaged users among viewers who sent comments on each of the posted images, based on the product attributes identified from each of all posted images posted by the poster within the prediction processing period and the language processing results (adjectives, adverbs) of the comments posted on each of the posted images.

The preferences of followers and engaged users may be estimated based on all images posted in the past by other users.

As a means for identifying a product, the program causes the processor 11 to
A means for identifying the type of product from the language processing results (nouns) of the poster's comments;
A means for selecting one convolutional neural network according to the identified product type from a plurality of convolutional neural networks prepared corresponding to a plurality of product types, respectively;
The posted image (target image) is applied to one convolutional neural network selected according to the type of product, thereby functioning as a means for identifying the product depicted in the posted image.

In this embodiment, a sales power score is obtained that quantifies the poster's ability to sell products through SNS based on the response score, consistency score, and customer service score. This can be used as basic information for objectively evaluating the ability of each poster. It is also possible to specifically grasp what is needed to increase the sales power score and what is lacking from the information that forms the basis of each score, the response score, consistency score, and customer service score, and it becomes possible to more specifically recognize what kind of SNS activity to increase. Furthermore, sales resulting from SNS activity can be predicted with high accuracy from the sales power score and sales performance.

Next, a processing procedure of the sales force prediction device 1 according to this embodiment will be described.
As shown in Fig. 6, for example, a sales prediction request is sent from the sales management device 2 or another terminal to the sales power prediction device 1 along with the SNS account of the salesperson (image poster) who is the subject of prediction. The sales power prediction device 1 requests the sales management device 2 to provide sales performance data. In response, the sales management device 2 replies to the sales power prediction device 1 with the sales performance data of the salesperson who is the subject of prediction. The sales performance data includes the product numbers sold within the prediction processing period, their sales prices, sales dates and times, etc. Note that the sales performance data is not limited to the prediction processing period, and may relate to all products sold by the poster in the past.

The sales force prediction device 1 also requests the sales web server 3 to provide the number of visitors who flowed to the sales web page via the page of the SNS account of the salesperson (image poster) in the SNS during the prediction processing period. In response, the sales web server 3 returns data on the number of visitors to the sales force prediction device 1.

Furthermore, the sales power prediction device 1 requests from the SNS server 4 a number of posted images posted by the poster using the SNS account during the prediction processing period, the profile of each follower who follows the poster at the time of prediction processing, the number of engagements (number of ratings, number of comments, number of saved posted images) for each posted image, the profile of each of a number of engaged users who performed a number of engagements (ratings, comments, saved posted images) for each posted image, comments for each posted image, and the number of replies by the poster to those comments and the contents of the replies. In response, the SNS server 4 replies to the sales power prediction device 1 with all posted images during the prediction processing period, the follower profiles at the current time, the number of engagements (number of ratings, number of comments, number of saved posted images) for each posted image for all posted images during the prediction processing period, the profiles of those engaged users, the comments, the number of replies, and the contents of the replies.

Next, processor 11 performs language analysis processing on the comment to extract nouns, adjectives, adverbs, etc. contained therein. Processor 11 estimates the product type (tops/bottoms, etc.) from the nouns resulting from the language analysis, selects a convolutional neural network corresponding to that type, and executes the convolutional neural network on the posted image to identify the product number of the product in the posted image.

In practice, as shown in FIG. 7, a separator 111 separates images from comments, the images are sent to an object extraction processing unit 112 and a posting category analysis unit 115, and the comments are sent to a language analysis unit 113.

The language analysis unit 113 performs text mining on the comments of the poster and the viewer to extract nouns, adjectives, and adverbs as parts of speech. If the comments contain pictures, signs, or symbols such as emoticons or stamps, they are extracted as adjectives. In most cases, nouns extracted from the image poster's comments represent the type of object of interest in the image. The nouns extracted from the image poster's comments (the codes representing them) are supplied to the product analysis unit 114, and the adjectives and adverbs (the codes representing them) are supplied to the preference analysis unit 116.

The object extraction processing unit 112 extracts and analyzes objects from the supplied image. This process typically applies a convolutional neural network to determine the type of object (major classification), the object's position (section within the frame), and the size of the object area (proportion of the number of pixels in the object area relative to the frame). For example, in the example of Figure 7, the image contains two objects, and the first object is of type "tops", is located in the upper central section, and has a size class of "medium". Similarly, the second object is of type "bottoms", is located in the lower central section, and has a size class of "medium". Based on this object information, a partial image is cut out. The cut out partial image is supplied to the product analysis unit 114.

In the product analysis unit 114, after performing preprocessing such as resolution normalization and contrast normalization on the partial image, the product number is identified by a convolutional neural network selected according to the product type estimated from the nouns extracted from the image poster's comments. In machine learning for creating a convolutional neural network, a huge number of actual teacher images are prepared and learning is repeated. As is well known, a convolutional neural network extracts features from an image by performing convolution using a filter on a two-dimensional image. A convolutional neural network is composed of a multi-layer network that repeats convolution and pooling. In the convolutional neural network, the coefficients (weights) of the filters that are effective for discrimination and that constitute the convolutional layer are trained using a large amount of data such as a large number of training images. The coefficients (weights) are obtained by performing learning to acquire invariance against various deformations by repeating convolution using a filter and pooling that summarizes responses in a certain area using a large amount of data. In addition, the discrimination performance of the convolutional neural network depends on the filters that constitute the convolutional layer. Filters are prepared for each type of object.

The posting category analysis unit 115 applies a convolutional neural network to the images supplied by the separator 111 to distinguish whether the posted image is an image of the product alone, an image that includes the product in the composition, and even more specifically, within the image category that includes the product in the composition, whether it is an image of a coordinated outfit in which the product is worn in combination with other products (clothing), an image of a model wearing the product in a relaxed pose in a landscape, or an image of a model wearing the product in a workplace, etc.

The preference analysis unit 116 reads from the storage device 17 the product attribute data corresponding to the product number identified by the product analysis unit 114, and estimates the user's preference trends by associating each item of the product attributes (color tone, product type, product category, design classification, etc.) with a preference polarity (positive (plus)/negative (minus)). The product category is associated with the purchasing tendency of the preference trend. When estimating the user's preference trends, tables for determining preference polarity from adjectives are created in advance for each generation and gender and stored in the storage device 14, and the adjectives extracted by the language analysis unit 113 are referenced in the table selected according to the age and gender in the user profile that posted the comment, to estimate the user's subjective preference polarity.

This taste tendency estimation process targets the current (at the time of prediction process) followers and engagement users of the target salesperson (image poster). The posts to be analyzed for the taste tendency of the follower and engagement user are not limited to the posts of the target salesperson (image poster), but should also include other posted images and comments posted by other salespersons at the same store to which the follower and engagement user are commenting, and it is preferable to apply this to as many posts as possible in order to increase the accuracy of the taste tendency. Then, the number of positive (plus) and negative (minus) points are counted for the taste polarity for each product attribute item that is estimated and collected from many posts and comments, and the overall taste polarity is determined depending on the number. The degree of preference may also be determined based on the ratio of the positive (plus) or negative (minus) points to the total number.

Next, the degree of preference consistency is calculated to determine the degree to which the preferences of followers and engaged users are consistent with the attributes of the product in the posted image (target image) that is the subject of the sales force prediction process. This process compares the preference polarity of each attribute item of followers and engaged users with the attributes of the product in the posted image, for example, and determines whether there is a match. For example, if the "color tone" item of the product attribute is red and the preference polarity for the color tone "red" is positive, the counter is incremented by one. If the color tone attribute of the product is red, but the preference polarity for the color tone red is negative or positive for other colors, the counter increment process is not executed. This counting process is performed to compare the preference trends of all followers and engaged users with all attribute items of the products in all posted images. The consistency of the preference trends of followers and engaged users with the attributes of the product in the posted image is calculated by dividing the count value by the total number of matches.

In addition, the degree of consistency of the target demographic that has been preset for the product in the posted image that is the subject of prediction processing, that is, the ratio of the number of followers and engaged users that match the target demographic to the total number of followers and engaged users, is calculated for the profile (age, gender) of the current (at the time of prediction processing) followers and engaged users of the target salesperson (image poster).

The consistency of the followers' and engaged users' preferences for the attributes of the product in the posted image obtained in this way, and the consistency of the followers' and engaged users for the target demographic of the product in the posted image are applied to a neural network to calculate a consistency score.

The consistency score can be increased by posting product images that match the attributes of followers and engaged users based on their preferences, and by posting product images that have a target demographic that matches the profiles of followers and engaged users.

It is preferable to compile the results of the aggregation of the preference trends of the salesperson (poster) and his/her posts among followers and engagement users, as well as the age and gender distribution of the followers and engagement users, and transmit this distribution together with a list of the product's target demographic and a consistency score to the sales management device 2 and user terminal 5, and provide feedback to the sales store and salesperson (poster) as reference material.

In addition, the number of engagements received for the latest one or several posted images that are the subject of the sales power prediction process, that is, the number of ratings (in principle the number of "likes", but the number of "dislikes" can also be included), the number of comments, and the number of saved images, as well as a product-specific score that is predefined for the product (product number) of the posted image, are applied to the neural network to calculate a reaction score for each posted image that indicates the degree of user reaction to each posted image. The reaction score increases as the number of "likes", the number of comments, and the number of saved images increase. The reaction score also increases by posting images of products with a high product-specific score and by increasing the number of posts.

It is preferable to transmit the number of "likes," the number of comments, the number of saved images, and the product-specific score, together with the consistency score, to the sales management device 2 and the user terminal 5, and provide feedback to the sales store or salesperson (poster) as reference material.

In addition, a customer service score that represents the customer service ability of the poster is calculated by applying to a neural network the number of all images posted by the salesperson (image poster) who is the target of the sales power prediction process during the prediction processing period, the number of inflows from those posts (total inflows), as well as the number of followers and following at the time of the prediction processing, the number of comments on all posted images during the prediction processing period, the number of replies to comments, the reply rate to comments, and the category scores of all posted images during the prediction processing period. The customer service score increases as the number of posted images, the number of inflows, the number of followers, the number of followings, the number of comments, the number of replies to comments, and the reply rate to comments increase, and also increases when the posted image is an outfit image or an image of the product being worn rather than an image of the product alone.

It is preferable to transmit the number of posted images, the number of visitors, the number of followers, the number of followings, the number of comments, the number of replies to comments, the reply rate to comments, and a list of posted image category scores to the sales management device 2 and the user terminal 5, and provide feedback to the sales store and sales staff (posters) as reference material.

In this way, the response score and consistency score calculated for each image posted by the salesperson (image poster) who is the target of the sales power prediction process during the prediction processing period, and the customer service score calculated comprehensively from the images posted by the salesperson (image poster) who is the target of the sales power prediction process during the prediction processing period, are applied to a neural network to calculate a sales power score that represents the ability of the salesperson (image poster) who is the target of the sales power prediction process to sell products through SNS at the time of the prediction processing.

The sales power score increases according to increases in each response score, consistency score, and customer service score. It is preferable to transmit a list of the sales power scores, response scores, consistency scores, and customer service scores for all sales staff (contributors) at the store to the sales management device 2 and the user terminal 5, and provide feedback to the sales store and sales staff (contributors) as reference material.

Furthermore, from the sales results, weekly sales results and their changes over time, monthly sales results and their changes over time, weekly sales points and their changes over time, monthly and weekly sales points and their changes over time, the product number distribution of sold products, and the distribution of types of sold products are compiled, and further characteristic values (median, maximum, minimum, etc.) of the sales price range are obtained, and in addition to various parameters derived from these sales results, the sales power score is applied to the neural network to predict the sales of products by the salesperson (poster) via SNS this week, this month, and further the next week, next month, next year, etc.

These sales forecast data are sent to the sales management device 2 along with the target salesperson's response score, consistency score, and customer service score. If necessary, they are also sent to the salesperson's terminal 5.

In this manner, according to this embodiment, SNS information is utilized to calculate the response score, consistency score, and customer service score, a sales power score is obtained that quantifies the poster's ability to sell products through SNS, and sales from SNS activities can be predicted with high accuracy from the sales power score and sales results. The sales power score and sales forecast can be used as basic information for objectively evaluating the ability of each poster to sell products through SNS. Furthermore, it is possible to specifically grasp what is needed and what is lacking in order to increase the sales power score from the response score, consistency score, and customer service score, and it becomes possible to more specifically recognize what kind of SNS activity to increase.

Although several embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the gist of the invention. These embodiments and their modifications are within the scope of the invention and its equivalents as set forth in the claims, as well as the scope and gist of the invention.

1...Sales force prediction device, 2...Sales management device, 3...Sales web server, 4...SNS server, 5-8...User terminals.

Claims (10)

An information processing device connected to an SNS server that provides an SNS via an Internet line,
A means for requesting and receiving from the SNS server a profile of each follower who follows a poster who posted a plurality of posted images, and a profile of each of a plurality of engaged users who performed a plurality of types of engagement (rating, commenting, saving the posted image) on each of the posted images;
A means for calculating a consistency score representing a consistency of the follower and the engaged user with a target demographic preset for a specific product in the posted image based on the follower profile and the engaged user profile;
An information processing device comprising: A means for requesting and receiving data related to the comments from the SNS server;
A means for estimating the preference tendency of the follower or the engagement user who sent the comment based on the attribute of the specific product and the language processing result of the comment;
The system further includes a means for calculating a consistency degree of the preference trends of the followers and the engagement users with respect to the attributes of the specific product,
The means for calculating the consistency score calculates the consistency score based on a consistency between the followers and the engagement users with a target group preset for the specific product, as well as a consistency between the taste tendencies of the followers and the engagement users with attributes of the specific product.
2. The information processing device according to claim 1 . A means for identifying a type of product in the posted image based on a result of language processing of the comment;
The information processing device according to claim 2 , further comprising: a means for identifying a product shown in the posted image by applying the posted image to a convolutional neural network selected according to the identified type of product. 2. The information processing apparatus according to claim 1 , wherein the means for calculating the consistency score comprises a neural network. A sales force prediction device connected via an internet line to a sales web server that sells products via a web page and an SNS server that provides an SNS,
A means for requesting and receiving a plurality of posted images posted by a poster who is a seller of the product from the SNS server;
A means for requesting and receiving from the sales web server the number of visitors who visited the web page via the SNS;
A means for calculating a customer service score representing the customer service ability of the poster based on the number of the posted images and the number of visitors;
An information processing device comprising: The method further includes a means for determining whether the posted image belongs to a product-only category or a category including the product in a composition by applying the posted image to a convolutional neural network,
The information processing device according to claim 5 , wherein the means for calculating the customer service score calculates the customer service score based on the number of posted images, the number of visitors, and a score assigned to a category of the posted images. The method further includes a means for requesting and receiving comments and replies to each of the posted images from the SNS server,
The information processing device according to claim 6, wherein the means for calculating the customer service score calculates the customer service score based on the number of followers and the response rate to the comments in addition to the number of posted images, the number of visitors, and the score assigned to the category of the posted images. 6. The information processing device according to claim 5 , wherein the means for calculating the customer service score is configured by a neural network. A computer connected to an SNS server that provides an SNS via an Internet line,
A means for requesting and receiving from the SNS server a profile of each follower who follows a poster who posted a plurality of posted images, and a profile of each of a plurality of engaged users who performed a plurality of types of engagement (rating, commenting, saving the posted image) on each of the posted images;
A program for functioning as a means for calculating a consistency score representing the degree of consistency of the follower and the engaged user with a pre-set target group for a specific product in the posted image based on the follower's profile and the engaged user's profile. A computer connected via an Internet line to a sales web server that sells products via a web page and an SNS server that provides an SNS,
A means for requesting and receiving a plurality of posted images posted by a poster who is a seller of the product from the SNS server;
A means for requesting and receiving from the sales web server the number of visitors who visited the web page via the SNS;
A program for functioning as a means for calculating a customer service score representing the customer service ability of the poster based on the number of posted images and the number of visitors.