Method and device for recommending sales product and predicting sales amount of influencer

Abstract

Provided are a method and device for recommending a sales product and predicting a sales amount of an influencer. The device for recommending a sales product and predicting a sales amount of an influencer according to an embodiment comprises: a data collection unit that collects a plurality of pieces of sales history data respectively corresponding to a plurality of influencers and a plurality of pieces of SNS posting data respectively corresponding to the plurality of influencers; a posting data processing unit that analyzes the plurality of pieces of SNS posting data; a data storage unit that stores the plurality of pieces of sales history data, a plurality of pieces of keyword data, and a plurality of pieces of posting quantitative data for the respective influencers; a training unit that trains a product recommendation model with the data stored in the data storage unit; and a sales amount prediction unit that predicts a sales amount of a recommended product, the sales amount being output from the product recommendation model in correspondence to a target influencer. The training unit trains a deep learning-based product prediction model which recommends a sales product to the influencer on the basis of the input data.

Description

Method and device for recommending products and predicting sales volume through influencers

[0001] The present invention relates to a method and device for recommending products and predicting sales volumes of recommended products. Specifically, the present invention relates to a method and device for recommending products to influencers, taking into account the influencer's sales history, relationships with other influencers, and posting information, and for providing a predicted sales volume of the recommended products.

[0002] The content described in this section merely provides background information for the present embodiment and does not constitute prior art.

[0003] This research was supported by the Ministry of SMEs and Startups, Small and Medium Business Technology Innovation Development (SME Accounting) Project [Small Business Owner-Social Seller Matching and Sales Platform, Cami Partner Center, Project Identification Number: 1425176712, Subproject Number: 00255887].

[0004] Influencer marketing is a marketing strategy that leverages influential social media influencers to promote products or services. Influencers often have large followings based on their expertise or interests in a specific field. Their content can inform consumers about products or services, influencing their purchasing decisions. Companies that sell or manufacture products can also effectively leverage influencer marketing to increase brand awareness, boost conversion rates, and strengthen customer loyalty.

[0005] Traditionally, influencer marketing involved companies selecting specific influencers and commissioning them to write advertisements or reviews. However, recent trends have seen influencers establish their own brands and sell products through group purchases, or collaborate with companies to develop new brands and products and then sell them through group purchases. Influencers' influence is maintained and expanded based on their followers (consumers). Consumers are more influenced by the expertise and relevance of the information provided by influencers. Consumers tend to trust, accept, and rely on information provided by influencers when they perceive them to be highly professional and relevant.

[0006] Therefore, in recent influencer marketing, selecting products that align with the influencer's expertise and background has become the most crucial factor for successful influencer marketing and group buying. Furthermore, when influencers directly promote group buying, they are responsible for managing inventory, requiring them to predict sales volumes and manage product production and inventory.

[0007] Accordingly, the applicant of the present invention proposes a method and device for recommending products for sale to an influencer by considering the influencer's sales history, relationships with other influencers, posting information of the influencer, etc., and predicting and providing the sales volume of the recommended product.

[0008] The purpose of the present invention is to provide a method for recommending a product suitable for sale to an influencer by considering the influencer's sales history, relationships with other influencers, posting information of the influencer, etc., and predicting and providing the sales volume of the recommended product.

[0009] In addition, an object of the present invention is to provide a device that can recommend a sales product suitable for an influencer by considering the influencer's sales history, relationships with other influencers, posting information of the influencer, etc., and predict and provide the sales volume of the recommended product.

[0010] The purposes of the present invention are not limited to those mentioned above. Other purposes and advantages of the present invention not mentioned above can be understood through the following description and will be more clearly understood through the embodiments of the present invention. Furthermore, it will be readily apparent that the purposes and advantages of the present invention can be realized by the means and combinations thereof set forth in the claims.

[0011] According to some embodiments of the present invention, a device for recommending products sold and predicting sales volume by an influencer comprises: a data collection unit for collecting a plurality of sales history data corresponding to each of a plurality of influencers and a plurality of SNS posting data corresponding to each of the plurality of influencers, wherein the plurality of sales history data includes a sale item, a sale date, and a sale quantity, and the SNS posting data includes a posting date, a posting content, and a feedback content; a posting data processing unit for analyzing the plurality of SNS posting data, wherein the posting data processing unit extracts advertisement posting data for advertising and selling a product from among the plurality of SNS posting data, identifies an advertisement target product from the advertisement posting data, and analyzes the advertisement posting data to generate a plurality of keyword data and a plurality of posting quantitative data; a data storage unit for storing the plurality of sales history data, the plurality of keyword data, and the plurality of posting quantitative data for each of a plurality of influencers; a learning unit for learning a product recommendation model using data stored in the data storage unit;And a sales volume prediction unit that predicts the sales volume of recommended products output from the product recommendation model in response to the target influencer, wherein the learning unit configures the plurality of sales history data as a first learning dataset to learn a first product rating output model, configures the plurality of keyword data as a second learning dataset to learn a second product rating output model, configures the plurality of posting quantitative data as a third learning dataset to learn a third product rating output model, stacks the first output data of the learned first product rating output model, the second output data of the learned second product rating output model, and the third output data of the learned third product rating output model to configure prediction model learning data, and trains a deep learning-based product prediction model that recommends sales products for the influencer based on the prediction model learning data, thereby constructing the product recommendation model composed of the first product rating output model, the second product rating output model, the third product rating output model, and the product prediction model, and the sales volume prediction unit is configured to predict the relationship between the target influencer and a neighboring influencer and the neighboring influencer. Considering the previous sales volume of the influencer's recommended product, the predicted sales volume of the target influencer for the recommended product is calculated.

[0012] According to some embodiments of the present invention, a method for recommending a product for sale and predicting a sales volume of an influencer comprises: a data collecting step of collecting a plurality of sales history data corresponding to each of a plurality of influencers and a plurality of SNS posting data corresponding to each of the plurality of influencers, wherein the plurality of sales history data includes a sale item, a sale date, and a sale quantity, and the SNS posting data includes a posting date, a posting content, and a feedback content; a step of analyzing the plurality of SNS posting data to generate a plurality of keyword data and a plurality of posting quantitative data, the step comprising extracting advertisement posting data for advertisement and sale of a product from among the plurality of SNS posting data, identifying an advertisement target product from the advertisement posting data, and analyzing the advertisement posting data to generate a plurality of keyword data and a plurality of posting quantitative data; a step of storing the plurality of sales history data, the plurality of keyword data, and the plurality of posting quantitative data for each of a plurality of influencers; A step of configuring the plurality of sales history data as a first learning dataset to train a first product rating output model, configuring the plurality of keyword data as a second learning dataset to train a second product rating output model, and configuring the plurality of posting quantitative data as a third learning dataset to train a third product rating output model; A step of stacking the first output data of the trained first product rating output model, the second output data of the trained second product rating output model, and the third output data of the trained third product rating output model to configure prediction model learning data; A step of training a deep learning-based product prediction model that recommends sales products to an influencer based on the prediction model learning data to build the product recommendation model, which is comprised of the first product rating output model, the second product rating output model, the third product rating output model, and the product prediction model;A step of outputting recommended products for a target influencer through the product recommendation model; and a step of calculating a predicted sales volume of the target influencer for the outputted recommended products, the step including calculating the predicted sales volume of the target influencer for the recommended products by taking into account the relationship between the target influencer and a neighboring influencer and the previous sales volume of the recommended products of the neighboring influencer.

[0013] A computer program according to some embodiments of the present invention is stored in a medium in combination with hardware to execute the method for recommending products sold by the influencer and predicting sales volume.

[0014] A method and device for recommending products and predicting sales volume of influencers according to some embodiments of the present invention can recommend products suitable for the influencer based on the influencer's sales history data and SNS posting data, and can support the influencer in selecting products of a brand that better matches his or her expertise, competitiveness, and suitability and in successfully conducting a market.

[0015] Additionally, according to some embodiments, a method and device for recommending products and predicting sales volumes for influencers can predict sales volumes of products recommended by the influencer, taking into account the influencer's relationships with neighboring influencers and their sales volumes. This can assist influencers and brands in developing appropriate market plans and managing product inventory.

[0016] Additionally, according to some embodiments, a method and device for recommending products and predicting sales volume for influencers can construct product rating output models based on sales history data, posting quantitative data, and keyword data, and then stack the data output from the constructed product rating output models to construct a product prediction model. In other words, by considering various information related to the influencer, more suitable product recommendations can be made for the influencer.

[0017] In addition, the method and device for recommending products and predicting sales volume of influencers according to some embodiments divides learning data into specific period units during the learning process and builds a product rating output model for each, and builds a product prediction model based on the result data, so that the product prediction model can proceed with learning by considering even time-series changes in input data.

[0018] In addition, a method and device for recommending products and predicting sales volume of influencers according to some embodiments can identify other influencers with a high relationship with the influencer, determine their sales volume by considering the sales characteristics of the products, and predict the sales volume of products recommended to the influencer.

[0019] In addition to the above-described contents, the specific effects of the present invention are described together with the specific matters for carrying out the invention below.

[0020] FIG. 1 is an exemplary diagram illustrating a brokerage system for matching influencers and brands according to some embodiments of the present invention.

[0021] FIG. 2 illustrates an exemplary seller search interface according to some embodiments of the present invention.

[0022] FIG. 3 illustrates an example of a product search interface according to some embodiments of the present invention.

[0023] FIG. 4 illustrates an example of a market management interface for managing multiple markets run on the seller side according to some embodiments of the present invention.

[0024] FIG. 5 is a block diagram illustrating the main configuration of a server according to some embodiments of the present invention.

[0025] FIG. 6 is an exemplary diagram illustrating a data processing process of a data collection unit, a posting data processing unit, and a data storage unit according to some embodiments of the present invention.

[0026] FIG. 7 illustrates sales history data according to some embodiments of the present invention.

[0027] FIG. 8 illustrates keyword data and posting quantitative data according to some embodiments of the present invention.

[0028] FIG. 9 is an exemplary diagram illustrating the operation process of a learning unit according to some embodiments of the present invention.

[0029] FIG. 10 illustrates an example of a first influencer-product rating matrix according to some embodiments of the present invention.

[0030] FIG. 11 illustrates an example of a first influencer similarity according to some embodiments of the present invention.

[0031] FIG. 12 illustrates an example process of applying multiple sliding windows according to some embodiments of the present invention.

[0032] FIG. 13 illustrates data stacked based on a first influencer, product rating stacking data of the first influencer, as an example, of the results of a plurality of product rating output models according to some embodiments of the present invention.

[0033] FIG. 14 is an exemplary diagram illustrating cross-validation according to some embodiments of the present invention.

[0034] FIG. 15 is an exemplary diagram illustrating a product prediction model according to some embodiments of the present invention.

[0035] FIG. 16 is an exemplary diagram illustrating the operation of a sales volume prediction unit according to some embodiments of the present invention.

[0036] Figure 17 is a flowchart of a method for recommending products and predicting sales volume by an influencer according to some embodiments of the present invention.

[0037] FIG. 18 is a diagram illustrating a hardware implementation of a device that performs a method for recommending products and predicting sales volume of influencers according to some embodiments of the present invention.

[0038] The terms and words used in this specification and claims should not be interpreted based on their general or dictionary meanings. In accordance with the principle that inventors can define the concepts of terms and words to best describe their inventions, they should be interpreted in a way that is consistent with the technical concept of the present invention. Furthermore, the embodiments described in this specification and the configurations depicted in the drawings are merely examples of how the present invention can be realized and do not fully represent the technical concept of the present invention. Therefore, it should be understood that various equivalents, modifications, and applicable examples may exist as of the time of filing.

[0039] The terms first, second, A, B, etc. used in this specification and claims may be used to describe various components, but the components should not be limited by these terms. These terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component. The term "and / or" includes any combination of a plurality of related listed items or any item among a plurality of related listed items.

[0040] The terminology used in this specification and claims is for the purpose of describing specific embodiments only and is not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. It should be understood that terms such as "comprise" or "have" in this application do not preclude the presence or addition of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification.

[0041] In addition, each configuration, process, procedure or method included in each embodiment of the present invention may be shared within a scope that is not technically inconsistent with each other.

[0042] Additionally, unless otherwise defined herein, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention pertains.

[0043] Terms defined in commonly used dictionaries should be interpreted to have a meaning consistent with their meaning in the context of the relevant technology, and will not be interpreted in an idealized or overly formal sense unless expressly defined in this application.

[0044]

[0045] Hereinafter, with reference to FIGS. 1 to 1, a method and device for recommending products and predicting sales volume by an influencer according to some embodiments of the present invention will be described.

[0046] Figure 1 is an exemplary diagram illustrating a brokerage system that matches influencers and brands according to some embodiments of the present invention. Figure 2 illustrates a seller search interface according to some embodiments of the present invention. Figure 3 illustrates a product search interface according to some embodiments of the present invention. Figure 4 illustrates a market management interface for managing multiple markets run by sellers according to some embodiments of the present invention.

[0047] A brokerage system (10) according to some embodiments of the present invention can support brokering and matching between influencers and brands.

[0048] Referring to FIG. 1, a brokerage system (10) according to an embodiment includes a server (100), a plurality of influencer terminals (200), a plurality of brand terminals (300), a plurality of SNS servers (400), and a plurality of market servers (500).

[0049] In the present invention, an influencer refers to a user who has a large number of followers on a social network service (hereinafter, "SNS"), including Facebook, Twitter, Instagram, YouTube, and internet blogs, and exerts influence on other users. In the present invention, an influencer may be an entity that sells a brand's products on behalf of the brand or sells products in collaboration with the brand, and may also be referred to as a "seller."

[0050] In the present invention, the "market" refers to the overall process by which matched brands and influencers sell products. For example, the "market status" refers to the status of product orders, cancellation management, return management, exchange management, and settlement for product sales.

[0051] Additionally, in the present invention, a brand may refer to an entity (company or individual) that develops, produces, and sells a product. Brands may need to connect with influencers to promote and market their products. Influencers may also need to be matched with brands that possess suitable products for group purchases, auctions, and other initiatives.

[0052] The server (100) can provide a brokerage service that matches multiple influencers with multiple brands and supports product sales and joint purchases through influencer marketing between the matched influencers and brands. Furthermore, the server (100) can easily monitor the progress of established markets between matched brands and influencers, and can provide influencers and brands with a user environment that allows for overall market management. Furthermore, the server (100) can recommend products from a brand to sell to influencers and provide estimated sales volumes for the recommended products, thereby supporting influencers in selecting products to sell and brands in managing product inventory.

[0053] The server (100) may be configured to exchange data with the server (100), multiple influencer terminals (200), multiple brand terminals (300), multiple SNS servers (400), and multiple market servers (500) via a network. Here, the network may include a network based on wired Internet technology, wireless Internet technology, and short-range communication technology. For example, the wired Internet technology may include at least one of a local area network (LAN) and a wide area network (WAN).

[0054] Wireless Internet technologies may include, for example, at least one of Wireless LAN (WLAN), Digital Living Network Alliance (DLNA), Wireless Broadband (Wibro), World Interoperability for Microwave Access (Wimax), High Speed ​​Downlink Packet Access (HSDPA), High Speed ​​Uplink Packet Access (HSUPA), IEEE 802.16, Long Term Evolution (LTE), Long Term Evolution-Advanced (LTE-A), Wireless Mobile Broadband Service (WMBS), and 5G NR (New Radio) technologies. However, the present embodiment is not limited thereto.

[0055] Short-range communication technologies may include, for example, at least one of Bluetooth, Radio Frequency Identification (RFID), Infrared Data Association (IrDA), Ultra-Wideband (UWB), ZigBee, Near Field Communication (NFC), Ultra Sound Communication (USC), Visible Light Communication (VLC), Wi-Fi, Wi-Fi Direct, and 5G NR (New Radio). However, the present embodiment is not limited thereto.

[0056] The server (100) can provide a user environment that can use a brokerage service that matches multiple influencers and multiple brands to multiple influencer terminals (200) and multiple brand terminals (300).

[0057] The server (100) can provide a service environment to users through a service application installed on each of the influencer terminal (200) and the brand terminal (300). Here, the service application may be a dedicated application for providing a service environment or a web browsing application for providing the service environment through a web page.

[0058] The influencer terminal (200) and the brand terminal (300) may be electronic devices capable of providing a user experience. For example, the influencer terminal (200) and the brand terminal (300) may be a smartphone, a type of portable terminal, but the present invention is not limited thereto. The influencer terminal (200) and the brand terminal (300) may include various types of electronic devices, such as a personal computer (PC), a laptop, a tablet, a mobile phone, a smartphone, or a wearable device (e.g., a watch-type terminal).

[0059] The plurality of SNS servers (400) may be servers that provide social network services (hereinafter, “SNS”), including Facebook, Twitter, Instagram, YouTube, and Internet blogs. Multiple influencers may be registered with SNSs provided by the plurality of SNS servers (400) and may post various information on these SNSs to deliver and provide various information to their followers.

[0060] The server (100) can collect information from multiple SNS servers (400) and provide a matching service that matches influencers with brands based on the collected information. The server (100) can provide an environment where users of the brokerage service can check information about multiple influencers based on the information collected from multiple SNS servers (400).

[0061] Multiple market servers (500) can provide a market service where products are sold and purchased online. The market service of multiple market servers (500) can be an open market, an e-commerce site open to both sellers and buyers. Brands or influencers can also participate in the market service, and sell manufactured products or products purchased through group purchases through the market service of the market server (500).

[0062] The server (100) can collect product-related information from multiple market servers (500). For example, price information related to a specific product can be collected. The server (100) can provide users of the brokerage service with a user environment that provides price information for products registered by the brand along with corresponding product information. This user environment can assist influencers in searching for products they wish to sell.

[0063] A brand can access the user environment provided by the server (100) through a brand terminal (300), and search for and select influencers that match the brand's image, product type, marketing direction, and purpose.

[0064] Figure 2 exemplarily illustrates a state in which Brand A utilizes a user interface to search for influencers (sellers). Referring to Figure 2, a seller search interface (I1) may be provided to the brand, allowing the brand to check the status of previous market progress, follower counts, etc., of multiple influencers. If the brand selects a specific influencer from the seller search interface (I1), a detailed information interface (I2) may be provided, providing detailed information about the selected influencer. Additional information may be provided, such as the influencer's main field, market progress with the influencer, and expected sales.

[0065] Influencers can access the user environment provided by the server (100) through the influencer terminal (200) and search for and select products of brands that are suitable for their expertise, competitiveness, and suitability.

[0066] Figure 3 exemplarily illustrates a state in which influencer A (seller A) utilizes a user interface to search for products to sell. Referring to Figure 3, the influencer can check information on multiple products registered by multiple brands through a product search interface (I3). The product search interface (I3) can provide price information (seller supply price, group purchase price) registered by the brand for each registered product, along with the price at which the product is sold on other open markets. Specifically, the seller supply price is the price the brand provides to the seller, and the group purchase price is the price at which the product is sold during the market process. The influencer can select a specific product and check detailed information through a detailed information interface (I4). Specifically, the detailed information interface (I4) can provide information such as the lowest online price, group purchase price, seller margin, the brand's group purchase history, and sample delivery conditions.

[0067] Influencers and brands can request market progress from their desired counterparts through the aforementioned search interfaces (I3, I1), and if mutual conditions are met, the market progress will be completed.

[0068] The server (100) can provide a market management interface (I5) as illustrated in FIG. 4. The market management interface (I5) can check the progress of product sales between influencers and brands that have completed a market, by influencer or brand. FIG. 4 shows a state in which the ongoing market status is checked based on Brand A. The progress of each influencer who has completed a market with Brand A, as well as related events, can be checked through the market management interface (I5).

[0069] According to an embodiment of the present invention, a server (100) can collect SNS posting data related to SNS activities of multiple influencers from multiple SNS servers (400), recommend sales products suitable for the influencers based on the SNS posting data, and predict sales volume for the recommended sales products. In addition, the server (100) can manage history data of products sold by influencers or brands matched through a brokerage service through a market server (500). Through data exchange with the market server (500), the server (100) can support sales management for products sold by influencers or brands, and collect sales history (sales products, sales volume, sales date, sales price, etc.). The server (100) can recommend sales products suitable for the influencers based on the sales history data, and predict sales volume for the recommended sales products.

[0070] These product recommendations and sales volume predictions can assist influencers in selecting more suitable and appropriate products by providing them with additional information to reference before selecting a specific product. In some examples, the server (100) may first display recommended products to influencers in the product search interface (I3) of FIG. 3, or provide additional indications indicating that the product is a recommended product, thereby facilitating the influencer's easy identification of the recommended product. Furthermore, the server (100) may further provide expected sales volumes for recommended products in the detailed interface (I4), thereby assisting influencers and brands in developing appropriate market plans and managing product inventory.

[0071]

[0072] Hereinafter, with reference to FIGS. 5 to 16, the main configuration of the server (100) and the process of recommending products for sale and predicting sales volume by an influencer performed in the server (100) will be described in more detail.

[0073]

[0074] FIG. 5 is a block diagram illustrating the main configuration of a server according to some embodiments of the present invention. FIG. 6 is an exemplary diagram illustrating a data processing process of a data collection unit, a posting data processing unit, and a data storage unit according to some embodiments of the present invention. FIG. 7 exemplarily illustrates sales history data according to some embodiments of the present invention. FIG. 8 exemplarily illustrates keyword data and posting quantitative data according to some embodiments of the present invention. FIG. 9 is an exemplary diagram illustrating the operation process of a learning unit according to some embodiments of the present invention. FIG. 10 exemplarily illustrates a first influencer-product rating matrix according to some embodiments of the present invention. FIG. 11 exemplarily illustrates a first influencer similarity according to some embodiments of the present invention. FIG. 12 exemplarily illustrates a process of applying multiple sliding windows according to some embodiments of the present invention. FIG. 13 exemplarily illustrates data obtained by stacking the results of a multiple product rating output model based on a first influencer and product rating stacking data of the first influencer according to some embodiments of the present invention. Figure 14 is an exemplary diagram illustrating cross-validation according to some embodiments of the present invention. Figure 15 is an exemplary diagram illustrating a product prediction model according to some embodiments of the present invention. Figure 16 is an exemplary diagram illustrating the operation of a sales volume prediction unit according to some embodiments of the present invention.

[0075]

[0076] Referring to FIG. 5, a server (100) according to some embodiments of the present invention includes a data collection unit (110), a posting data processing unit (120), a data storage unit (130), a learning unit (140), and a sales volume prediction unit (150).

[0077] The server (100) according to the embodiments may be entirely hardware, or may have aspects that are partially hardware and partially software. For example, the server (100) and each component included therein in the present specification are intended to refer to a combination of hardware and software driven by the hardware. The hardware may be a data processing device including a Central Processing Unit (CPU) or other processor. Furthermore, the software driven by the hardware may refer to a running process, object, executable, thread of execution, program, etc.

[0078] In addition, although the data collection unit (110), posting data processing unit (120), data storage unit (130), learning unit (140), and sales volume prediction unit (150) in FIG. 5 are depicted as distinct blocks, each unit is not necessarily intended to refer to a physically distinct, separate component. Each unit constituting the system or terminal is merely functionally distinguished by the operations performed by the corresponding device, and some or all of them may be integrated into the same device, one or more may be implemented as a separate device that is physically distinct from other units, and may be components that are communicatively connected to each other under a distributed computing environment.

[0079]

[0080] Referring to FIG. 6, the data collection unit (110) can collect data from multiple SNS servers (400) and multiple market servers (500). Specifically, the data collection unit (110) can crawl SNS posting data corresponding to each of multiple influencers from multiple SNS servers (400) via a network. In the example of FIG. 6, when a first influencer is using a first SNS server (400A) and a second SNS server (400B), the data collection unit (110) can collect posting data from each of the first SNS server (400A) and the second SNS server (400B) used by the first influencer.

[0081] The data collection unit (110) may collect SNS posting data by crawling data posted on the SNS accounts of multiple influencers at regular intervals, but embodiments of the present invention are not limited thereto. In some embodiments, the data collection unit (110) may collect SNS posting data of multiple influencers in response to a command from an administrator of the server (100). If an influencer uploads multiple SNS postings during a regular period, the data collection unit (110) may collect data for each of the multiple SNS postings.

[0082] The SNS posting data collected by the data collection unit (110) may include at least the posting date, posting content, and feedback. Here, the SNS posting data may be, but is not limited to, advertising posts for product advertisements or sales. The posting content may be data crawled from the text included in the title and body of a post uploaded by an influencer. Additionally, the posting content may include at least one image and video included in the post. The feedback content may be information related to the response of at least one of a follower, the influencer, and another influencer in response to the post. In some embodiments, the feedback content may include positive response information (e.g., "like"), negative response information (e.g., "dislike"), and reply information for the post. The reply information may be data crawled from comments written by at least one of a follower, the influencer, and another influencer in response to the post.

[0083] The SNS posting data collected by the data collection unit (110) can be provided to the posting data processing unit (120) described below, and preprocessing and data conversion for configuring input data or learning data can be performed in the posting data processing unit (120).

[0084] In addition, the data collection unit (110) can collect sales history data that can confirm the status of products sold by multiple influencers from multiple open market servers (500). The data collection unit (110) can collect sales history data according to a certain cycle, and can collect product sales history for each influencer within a period corresponding to the certain cycle. In the example of FIG. 6, if a first influencer sells products on a first open market server (500A) and a second open market server (500B) during the corresponding period, the data collection unit (110) can collect sales history data from the first open market server (500A) and the second open market server (500B), respectively.

[0085] Here, sales history data may include at least the items sold, classification information for the items sold, the number of units sold, and the date of sale. Furthermore, sales history data may include the selling price of the items sold. Here, classification information refers to defining multiple items into specific categories based on preset criteria. For example, refrigerators and air fryers may be classified as kitchen appliances, while multivitamins and probiotics may be classified as health foods. Sales history data collected by the data collection unit (110) may be provided to the data storage unit (130) and stored in association with influencer accounts. Data stored in the data storage unit (130) may be collated and stored based on classification information.

[0086] The posting data processing unit (120) can analyze SNS posting data to determine whether the posting is for the purpose of advertising or selling a product. In other words, it can identify whether the posting is for the purpose of advertising or selling a product, rather than a posting related to the influencer's general daily life. Furthermore, the posting data processing unit (120) can extract advertising post data from multiple SNS postings. Furthermore, the posting data processing unit (120) can identify the product being advertised from the advertising post data and identify product classification information. Furthermore, the posting data processing unit (120) can analyze multiple SNS posting data to extract advertising post data, identify the product being advertised from the extracted advertising post data, and analyze the advertising post data to generate multiple keyword data and multiple quantitative posting data.

[0087] Here, the keyword data may be derived from a qualitative analysis of an advertisement posting. Specifically, the posting data processing unit (120) can analyze the content of an advertisement posting to identify key keywords and count the frequency of use of the identified key keywords. In other words, the keyword data may consist of key keywords identified in the content posted by the influencer and their frequency of use, and the identified key keywords and frequency of use may be mapped to the posting date.

[0088] Posting quantitative data may be derived from quantitative analysis of advertising posts. The posting data processing unit (120) may analyze the content and feedback of a post to generate posting quantitative data that includes quantitative information related to the format of the post and quantitative information related to follower responses. In some embodiments, the posting quantitative data may include the number of positive responses, the total number of comments, the number of comments written by the influencer, the number of images, and the number of videos.

[0089] The posting data processing unit (120) can provide the generated plurality of keyword data and the plurality of posting quantitative data to the data storage unit (130).

[0090] The data storage unit (130) can store multiple sales history data, multiple keyword data, and multiple posting quantitative data corresponding to multiple influencer accounts. The data storage unit (130) stores the sales history data corresponding to each influencer account, and sequentially stores the sales history data by sorting it by sales date. For example, the data storage unit (130) can sort the stored data by sales month and store and manage data corresponding to the same month in an integrated manner. Figure 7 illustrates an exemplary state of sales history data stored in the data storage unit (130). Referring to Figure 7, the sales history data can be seen to be in a state where the quantity of products sold by multiple influencers is sorted by sales month. In other words, the product sales quantity according to the classification information of the sold product is organized by sales month.

[0091] The data storage unit (130) stores keyword data and posting quantitative data corresponding to influencer accounts. The keyword data and posting quantitative data can be sequentially stored by sorting them based on the posting date. For example, the data storage unit (130) can sort the stored data based on the posting month and store and manage data corresponding to the same month in an integrated manner.

[0092] FIG. 8 (a) exemplarily illustrates keyword data stored in the data storage unit (130), and FIG. 8 (b) exemplarily illustrates posting quantitative data stored in the data storage unit (130). Referring to FIG. 8 (a), it can be seen that the target product (health food) advertised and sold by the first influencer, the main keywords used in the advertising posting, and the frequency of the main keywords are sorted based on the posting month. In addition, referring to FIG. 8 (b), it can be seen that the target product (health food) advertised and sold by the second influencer, the number of likes, comments, replies, influencer replies, posting number, image number, and video number for the advertising posting are sorted based on the posting month.

[0093] The learning unit (140) can perform learning on a product recommendation model (RM) based on data stored in the data storage unit (130). Here, the product recommendation model (RM) can include a deep learning-based product prediction model (CM) trained to recommend products by receiving a plurality of product rating output models (CF) and result values ​​output from the product rating output models (CF).

[0094] Referring to FIG. 9, the learning unit (140) may select sales history data as a first learning data set, keyword data as a second learning data set, and posting quantitative data as a third learning data set, and may construct a first product rating output model (CF1), a second product rating output model (CF2), and a third product rating output model (CF3) corresponding to the first, second, and third learning data sets, respectively. The learning unit (140) may compile data output from the first product rating output model (CF1), the second product rating output model (CF2), and the third product rating output model (CF3) to construct new input data, and may train a product prediction model (CM) using the constructed input data. The trained product recommendation model (RM) may recommend sales products suitable for the target influencer to sell by considering the sales history data, keyword data, and posting quantitative data of the target influencer.

[0095] The learning unit (140) constructs a first influencer-product rating matrix (M1) based on sales history data, and can calculate a first influencer similarity (S1) based on the sales history data. The first influencer-product rating matrix (M1) may be a matrix illustrating the relationship between each influencer and each product, as illustrated in FIG. 10 . Here, the rating, which is the value of each matrix entry, may be determined based on the sales volume of the corresponding product by each influencer. For example, the higher the influencer's sales volume for the corresponding product, the higher the rating for the corresponding product. The first influencer similarity matrix (S1) corresponds to a matrix that determines the similarity between multiple influencers by analyzing sales history data. In an exemplary embodiment, influencers who sell the same or similar products and have similar sales volumes may be determined to have a high similarity value in the first influencer similarity matrix (S1). As shown in Figure 11, it can be seen that the similarity between multiple influencers is calculated based on sales history data.

[0096] The learning unit (140) can construct a first product rating output model (CF1) that outputs an expected rating of a product for each influencer based on the configured first influencer-product rating matrix (M1) and the first influencer similarity matrix (S1). Here, the learning unit (140) can further consider the average rating of each influencer and the average rating of other influencers to generate an expected rating (P) for the i-th product of influencer a. ai ) can be calculated as in mathematical equation 1.

[0097]

[0098] [Mathematical Formula 1]

[0099]

[0100]

[0101] (where n: number of influencers, P ai : Influencer A's predicted rating for product i, : The average rating of influencer a in the first influencer-product rating matrix (M1), W au : The similarity between influencer a and influencer u in the first influencer similarity matrix (S1), r ui : In the first influencer-product rating matrix (M1), the rating of influencer u for product i, : The average rating of influencer u in the first influencer-product rating matrix (M1)

[0102]

[0103] The learning unit (140) may construct a second influencer-product rating matrix (M2) based on keyword data, and may calculate a second influencer similarity (S2) based on the keyword data. The second influencer-product rating matrix (M2) may be a matrix illustrating the relationship between each influencer and each product. Here, the rating, which is the value of each item in the matrix, may be determined based on the keywords and the frequency of the keywords used by each influencer to promote and advertise the product. For example, if an influencer uses various keywords for the product and uses them frequently, the rating for the product may be determined to be high. The second influencer similarity matrix (S2) corresponds to a matrix that determines the similarity between multiple influencers by analyzing keyword data. In an exemplary embodiment, an influencer who sells the same or similar products and exhibits similar keywords and similar keyword frequencies may be determined to have a high similarity value in the second influencer similarity matrix (S2).

[0104] The learning unit (140) can construct a second product rating output model (CF2) that outputs the predicted rating of the product for each influencer based on the constructed second influencer-product rating matrix (M2) and the second influencer similarity matrix (S2). Here, the learning unit (140) further considers the average rating of each influencer and the average rating of other influencers to generate the predicted rating (for influencer a's product i) ) can be calculated as in mathematical equation 2.

[0105]

[0106] [Equation 2]

[0107]

[0108]

[0109] (where n: number of influencers, : Influencer A's predicted rating for product i, : The average rating of influencer a in the second influencer-product rating matrix (M2), : Similarity between influencer a and influencer u in the second influencer similarity matrix (S2), : In the second influencer-product rating matrix (M2), the rating of influencer u for product i, : The average rating of influencer u in the second influencer-product rating matrix (M2)

[0110]

[0111] The learning unit (140) constructs a third influencer-product rating matrix (M3) based on the quantitative posting data, and can calculate the third influencer similarity (S3) based on the quantitative posting data. The third influencer-product rating matrix (M3) may be a matrix illustrating the relationship between each influencer and each product. Here, the rating, which is the value of each matrix item, may be determined based on quantitative response data for each influencer's posting related to the product. For example, if a posting related to the product has a large number of likes, comments, images, and videos, the rating for the product may be determined to be high. The third influencer similarity matrix (S3) corresponds to a matrix that determines the similarity between multiple influencers by analyzing the quantitative posting data. In an exemplary embodiment, influencers who sell the same or similar products and have similar response data may be determined to have a high similarity value in the third influencer similarity matrix (S3).

[0112] The learning unit (140) can construct a third product rating output model (CF3) that outputs the predicted rating of the product for each influencer based on the constructed third influencer-product rating matrix (M3) and the third influencer similarity matrix (S3). Here, the learning unit (140) further considers the average rating of each influencer and the average rating of other influencers to generate the predicted rating for the i-th product of influencer a ( ) can be calculated as in mathematical formula 3.

[0113]

[0114] [Equation 3]

[0115]

[0116] (where n: number of influencers, : Influencer A's predicted rating for product i, : The average rating of influencer a in the third influencer-product rating matrix (M3), : Similarity between influencer a and influencer u in the third influencer similarity matrix (S3), : In the third influencer-product rating matrix (M3), the rating of influencer u for product i, : The average rating of influencer u in the third influencer-product rating matrix (M3)

[0117]

[0118] The data derived from the learning process of the learning unit (140) described above, for example, the first, second, and third influencer similarity matrices (S1, S2, S3), can be stored in the data storage unit (140) and can be utilized in the subsequent sales volume prediction process.

[0119] In some embodiments, the learning unit (140) divides multiple learning data sets into time periods using a sliding window, and constructs a product rating output model using each sub-learning data set divided into each time period. The process of constructing the above-described product rating output model can be constructed based on each sub-learning data set divided into each time period.

[0120] The learning unit (140) may include a sliding window (SW) with a preset period defined, and sales history data, keyword data, and posting quantitative data may be divided into units of a certain period through the sliding window (SW).

[0121] For example, referring to FIG. 12, the learning unit (140) may generate multiple first sub-learning data sets by segmenting sales history data through a sliding window, multiple second sub-learning data sets by segmenting keyword data through a sliding window, and multiple third sub-learning data sets by segmenting posting quantitative data through a sliding window. For example, the sliding window (SW) may be configured to extract data on a monthly basis, but is not limited thereto.

[0122] The learning unit (140) can generate multiple first product rating output models through multiple first sub-learning data sets generated, multiple second product rating output models through multiple second sub-learning data sets generated, and multiple third product rating output models through multiple third sub-learning data sets generated.

[0123] A plurality of first product rating output models may construct a first influencer-product rating matrix based on corresponding first sub-learning data sets, calculate a first influencer similarity based on the corresponding first sub-learning data sets, and output an expected rating of the product for each influencer based on the first influencer-product rating matrix and the first influencer similarity. A plurality of second product rating output models may construct a second influencer-product rating matrix based on corresponding second sub-learning data sets, calculate a second influencer similarity based on the corresponding second sub-learning data sets, and output an expected rating of the product for each influencer based on the second influencer-product rating matrix and the second influencer similarity. A plurality of third product rating output models can construct a third influencer-product rating matrix based on a corresponding third sub-learning data set, calculate third influencer similarity based on the corresponding third sub-learning data set, and output an expected rating of a product for each influencer based on the third influencer-product rating matrix and the third influencer similarity.

[0124] Multiple first product rating output models configured through a sliding window (SW) can analyze sales history data for multiple time periods and output predicted ratings for multiple influencers' products. In other words, the predicted ratings output from multiple first product rating output models correspond to data that can confirm time-series changes according to sales history data. Multiple second product rating output models configured through a sliding window (SW) can analyze keyword data for multiple time periods and output predicted ratings for multiple influencers' products. In other words, the predicted ratings output from multiple second product rating output models correspond to data that can confirm time-series changes according to keyword data. Multiple third product rating output models configured through a sliding window (SW) can analyze posting quantitative data for multiple time periods and output predicted ratings for multiple influencers' products. In other words, the predicted ratings output from multiple first product rating output models correspond to data that can confirm time-series changes according to posting quantitative data.

[0125] In the embodiments described below, the output values ​​from each of the first, second, and third product rating output models can be combined and utilized as input data for a product prediction model. These output values ​​correspond to data generated to reflect temporal changes. Therefore, by combining the output values ​​to form input data and using this to train a product prediction model, a product prediction model can be constructed that performs product predictions that also consider the temporal characteristics of the input data. In other words, product predictions that take temporal changes into account can be supported, and more suitable and accurate product recommendations can be made to users.

[0126] The learning unit (140) can collate and stack the predicted ratings, i.e., output data, for each of multiple products of multiple influencers from the constructed multiple product rating output models. Fig. 13 exemplarily illustrates a state in which the predicted ratings for each of multiple products of a first influencer are collated from the outputs of multiple product rating output models. Hereinafter, data collated in this manner is defined as influencer product rating stacking data.

[0127] The learning unit (140) can configure the collected output data as prediction model learning data for learning and building a product prediction model. That is, the prediction model learning data may be in a state in which product rating stacking data of each of multiple influencers is collected. In some embodiments, the learning unit (140) can divide the prediction model learning data into learning data and verification data, and can apply cross-validation, which cross-changes the learning data and verification data to learn the product prediction model, thereby building the product prediction model.

[0128] Referring to FIG. 14, the prediction model learning data may include first product rating stacking data corresponding to the first influencer and fifth product rating stacking data corresponding to the fifth influencer. For example, the ratio of learning data to verification data may be 4:1. The learning unit (140) may construct a learning data set by cross-referencing the types and orders of four product rating stacking data applied to the learning data and one product rating stacking data applied to the verification data, and may perform cross-validation to perform all learning through the constructed multiple learning data sets, thereby constructing a product prediction model.

[0129] Applying cross-validation prevents overfitting and data bias on specific datasets, enabling the construction of generalized models. Furthermore, by utilizing all datasets for training, accuracy can be improved and underfitting due to insufficient data can be prevented.

[0130] The learning unit (140) can learn and build a product prediction model that recommends products for sale to a specific influencer using this prediction model learning data. Specifically, the learning unit (140) can learn and build a product prediction model so that the product prediction model recommends products for sale in response to the product rating stacking data of a specific influencer. That is, as illustrated in FIG. 15, the product prediction model (CM) can recommend products for sale by a specific influencer in response to the product rating stacking data of that influencer.

[0131] In some embodiments, these product prediction models can be implemented using a deep learning module. The deep learning module can derive products using an artificial neural network trained on big data. The deep learning module can perform artificial neural network training using mapping data for separate parameters derived from the input data. The deep learning module can perform machine learning on the parameters input as learning factors.

[0132] Deep learning modules can utilize a variety of well-known deep learning architectures. For example, deep learning modules can utilize structures such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), deep belief networks (DBNs), and graph neural networks (GNNs).

[0133] Specifically, CNN (Convolutional Neural Network) is a model that simulates the function of the human brain, based on the assumption that when a person recognizes an object, he or she extracts the basic features of the object, performs complex calculations in the brain, and then recognizes the object based on the results.

[0134] Recurrent Neural Networks (RNNs) are widely used in natural language processing and other fields. They are effective at processing time-series data, which changes over time. Layers can be stacked at each moment to form an artificial neural network structure. In this case, when a product rating output model is built using sub-training data sets time-series-separated through sliding windows, RNNs may be more suitable for analyzing time-series relationships.

[0135] A DBN (Deep Belief Network) is a deep learning structure constructed by stacking multiple layers of Restricted Boltzman Machines (RBMs), a deep learning technique. By repeatedly training RBMs (Restricted Boltzman Machines), a certain number of layers can be created, creating a DBN (Deep Belief Network) with that number of layers.

[0136] GNN (Graphic Neural Network, hereinafter referred to as GNN) refers to an artificial neural network structure implemented by deriving similarities and feature points between modeled data using modeled data modeled based on data mapped between specific parameters. In other words, the product prediction model used in the present invention can utilize various deep learning modules depending on the type of input data. In addition, the product prediction model can additionally perform an operation of applying the input data to the deep learning module through a preprocessing process of parameterizing and processing the input data.

[0137] Meanwhile, artificial neural network training in the deep learning module can be achieved by adjusting the weights of connections between nodes (and, if necessary, adjusting bias values) to produce the desired output for a given input. Furthermore, artificial neural networks can continuously update their weight values ​​through learning. Furthermore, methods such as backpropagation can be used for artificial neural network training.

[0138] Meanwhile, the deep learning module can perform machine learning-based improvement process recommendations using modeling data on derived parameters as input data. Both semi-supervised and supervised learning can be used as machine learning methods for artificial neural networks. Furthermore, the deep learning module can be controlled to automatically update the artificial neural network structure for outputting products after learning, depending on the settings.

[0139] Additionally, the deep learning module includes an input layer (input) that uses product rating stacking data of a specific influencer as an input node, an output layer (output) that uses a sold product as an output node, and at least one hidden layer positioned between the input layer and the output layer.

[0140] Here, weights can be set for the edges connecting the nodes of each layer. These weights or the presence or absence of edges can be added, removed, or updated during the learning process. Therefore, the weights of the nodes and edges positioned between multiple input nodes and output nodes can be updated during the learning process. In other words, before the deep learning module performs learning, all nodes and edges can be set to initial values. However, as information is accumulated and input, the weights of the nodes and edges change, and during this process, a matching can be achieved between the parameters input as learning factors (i.e., product rating stacking data of a specific influencer) and the values ​​assigned to the output nodes (i.e., the products for sale).

[0141] Additionally, the weights of the nodes and edges between the input and output nodes that make up the deep learning module can be updated through the deep learning module's learning process. Furthermore, the parameters output from the deep learning module can be expanded to include various data beyond just products.

[0142] As described above, the learning unit (140) can configure a product recommendation model (RM) including multiple product rating output models (CF) and product prediction models (CM). The product recommendation model (RM) can recommend products suitable for sale by each influencer. The product recommendation model (RM) upon completion of training is illustrated as being included in the learning unit (140), but the embodiments of the present invention are not limited thereto.

[0143] When an influencer requiring product recommendation and sales volume prediction is defined as a target influencer, a product recommendation model (RM) can recommend a product suitable for the target influencer, and a sales volume prediction unit (150) can predict the sales volume for a product recommended by the product recommendation model (RM).

[0144] Referring to Figure 16, the sales volume prediction unit (150) can predict sales volume for products recommended by a product recommendation model (RM). This sales volume prediction can be determined by considering the similarity between the target influencer and other influencers and the past sales volume of multiple influencers for the product in question. The sales volume prediction unit (150) can calculate the predicted sales volume for a product (product j) recommended by the target influencer (influencer i) as shown in Mathematical Expression 4 below.

[0145]

[0146] [Equation 4]

[0147]

[0148] (Pr ij : Predicted sales volume for product j of influencer i, influencer i is the target influencer, product j is the recommended product, W in : The similarity between influencer i and neighboring influencer n, r is the number of neighboring influencers considered in calculating predicted sales, which is a natural number greater than or equal to 1, n is a natural number between 1 and r, and S nj : Past sales volume of product j of neighboring influencer n)

[0149]

[0150] Specifically, the sales volume prediction unit (150) can select a plurality of neighboring influencers having a high similarity to the target influencer. The sales volume prediction unit (150) can determine a plurality of neighboring influencers having a high similarity to the target influencer by considering the first influencer similarity matrix (S1), the second influencer similarity matrix (S2), and the third influencer similarity matrix (S3). For example, the sales volume prediction unit (150) can calculate an influencer similarity average matrix using the first influencer similarity matrix (S1), the second influencer similarity matrix (S2), and the third influencer similarity matrix (S3). The sales volume prediction unit (150) can determine the top r neighboring influencers having a high similarity to the target influencer by using the influencer similarity average matrix. Here, r may be a natural number greater than or equal to 1 and may be a preset reference value. For example, the sales volume prediction unit (150) can select the top 10 neighboring influencers with high similarity to the target influencer using the influencer similarity average matrix, but the embodiment of the present invention is not limited thereto.

[0151] The sales volume prediction unit (150) can extract the sales volume of recommended products sold in the past by neighboring influencers with high similarity. The past sales volume of a neighboring influencer can refer to the number of units sold during a reference period. The sales volume prediction unit (150) can define a reference period by considering the average market cycle of the recommended products and determine the past sales volume during the reference period.

[0152] Here, the average market cycle may refer to the cycle in which a market is formed between an influencer and a brand and sales of the corresponding product are performed. The sales volume prediction unit (150) may analyze sales history data to extract average market cycle data for each product and identify the average market cycle of the recommended product. The sales volume prediction unit (150) may determine a reference period considering the past and current sales volume of the recommended product based on the average market cycle. Here, the sales volume prediction unit (150) may determine the reference period as an average of four markets conducted. In other words, the number of days obtained by multiplying the average market cycle by 4 may be calculated as the reference period. For example, if the average market cycle of product j is 95 days, the reference period is 380 days, which is 95 days multiplied by 4. The sales volume prediction unit (150) may calculate the sales volume of recommended products sold by neighboring influencers from the present time when sales volume is being predicted to before the reference period. The sales volume prediction unit (150) can predict the product sales volume of the influencer by taking the average of the product sales volume of the recommended product sold by the neighboring influencer during the reference period and the similarity with the neighboring influencer.

[0153] In some embodiments, if an influencer has a history of selling the recommended product, that influencer may also be included among your neighbors. If the similarity values ​​of neighboring influencers are normalized so that 1 is the maximum and 0 is the minimum, then the influencer's similarity to you may be set to the maximum value of 1.

[0154] The product sales volume predicted by the sales volume prediction unit (150) can be provided to the influencer along with the recommended product. The influencer can determine which recommended product to market based on the recommended product and its expected sales volume. Furthermore, the influencer can plan and manage inventory for the market, taking into account the expected sales volume.

[0155]

[0156] Below, a method for recommending products and predicting sales volume for influencers according to an embodiment of the present invention will be described. The method for recommending products and predicting sales volume for influencers according to the following embodiment is performed by the server (100) described above. Reference may be made to FIGS. 1 through 16 and their related descriptions.

[0157] Figure 17 is a flowchart of a method for recommending products and predicting sales volume by an influencer according to some embodiments of the present invention.

[0158] Referring to FIG. 17, a method for recommending products sold and predicting sales volume by an influencer according to some embodiments of the present invention comprises: a data collecting step (S100) of collecting a plurality of sales history data corresponding to each of a plurality of influencers and a plurality of SNS posting data corresponding to each of the plurality of influencers; a step (S110) of analyzing the plurality of SNS posting data to generate a plurality of keyword data and a plurality of posting quantitative data; a step (S120) of storing the plurality of sales history data, the plurality of keyword data, and the plurality of posting quantitative data for each of the plurality of influencers; a step (S130) of configuring the plurality of sales history data as a first learning dataset to train a first product rating output model, configuring the plurality of keyword data as a second learning dataset to train a second product rating output model, and configuring the plurality of posting quantitative data as a third learning dataset to train a third product rating output model; and a step (S130) of stacking the first output data of the trained first product rating output model, the second output data of the trained second product rating output model, and the third output data of the trained third product rating output model to train a prediction model. The method includes a step of configuring data (S140), a step of learning a deep learning-based product prediction model that recommends products for sale to an influencer based on the prediction model learning data, and a step of building the product recommendation model composed of the first product rating output model, the second product rating output model, the third product rating output model, and the product prediction model (S150), a step of outputting recommended products for the target influencer through the product recommendation model (S160), and a step of calculating the predicted sales volume of the target influencer for the output recommended products (S170).

[0159] In some embodiments, step (S100) may include the plurality of sales history data including sales items, sales dates, and sales quantities, and the SNS posting data may include posting dates, posting content, and feedback content.

[0160] In some embodiments, step (S110) may include extracting advertisement posting data for advertising and selling a product from among the plurality of SNS posting data, identifying an advertised product from the advertisement posting data, and analyzing the advertisement posting data to generate a plurality of keyword data and a plurality of posting quantitative data.

[0161] In some embodiments, step (S170) may include calculating a predicted sales volume of the target influencer for the recommended product by considering the relationship between the target influencer and a neighboring influencer and the previous sales volume of the recommended product of the neighboring influencer.

[0162] In some embodiments, step (S100) collects SNS posting data by crawling data posted on SNS accounts of multiple influencers at regular intervals, wherein the posting content is data crawled from text included in the title and body of a post uploaded by a corresponding influencer and includes at least one image and video included in the post, and the feedback content may be information related to a reaction of at least one of a follower, the influencer, and another influencer in response to the post.

[0163] In some embodiments, the plurality of keyword data may be data that identifies main keywords of the SNS posting data and counts the frequency of the identified main keywords, and the posting quantitative data may be data that quantitatively analyzes the follower's response to the corresponding SNS posting data, the relationship with the follower, the number of postings, and the posting format.

[0164] In some embodiments, step (S130) constructs a first influencer-product rating matrix based on the sales history data, calculates a first influencer similarity based on the sales history data, constructs a first product rating output model that outputs an expected rating of a product for each influencer based on the first influencer-product rating matrix and the first influencer similarity, constructs a second influencer-product rating matrix based on the keyword data, calculates a second influencer similarity based on the sales history data, constructs a second product rating output model that outputs an expected rating of a product for each influencer based on the second influencer-product rating matrix and the second influencer similarity, constructs a third influencer-product rating matrix based on the posting quantitative data, calculates a third influencer similarity based on the sales history data, and constructs a third influencer-product rating matrix based on the third influencer-product rating matrix and the third influencer similarity. This may include building a third product rating output model that outputs predicted product ratings for each influencer.

[0165] In some embodiments, step (S130) may include constructing the first product rating output model, the second product rating output model, and the third product rating output model, respectively, by further considering the average rating of each influencer and the average rating of other influencers.

[0166] In some embodiments, step (S130) may include performing cross-validation, which divides the prediction model learning data into learning data and verification data, and cross-changes the learning data and the verification data to perform learning on the product prediction model.

[0167] In some embodiments, step (S170) may include identifying the neighboring influencer among the plurality of influencers based on the first influencer similarity, the second influencer similarity, and the third influencer similarity.

[0168] In some embodiments, step (S170) may include calculating an average market cycle of the recommended product, determining a reference period for considering the previous sales volume of the recommended product based on the calculated average market cycle, and calculating a predicted sales volume of the target influencer for the recommended product based on the previous sales volume of the recommended product previously sold by the neighboring influencer during the determined reference period.

[0169] A method and device for recommending products and predicting sales volume of influencers according to some embodiments of the present invention can recommend products suitable for the influencer based on the influencer's sales history data and SNS posting data, and can support the influencer in selecting products of a brand that better matches his or her expertise, competitiveness, and suitability and in successfully conducting a market.

[0170] Additionally, according to some embodiments, a method and device for recommending products and predicting sales volumes for influencers can predict sales volumes of products recommended by the influencer, taking into account the influencer's relationships with neighboring influencers and their sales volumes. This can assist influencers and brands in developing appropriate market plans and managing product inventory.

[0171] Additionally, according to some embodiments, a method and device for recommending products and predicting sales volume for influencers can construct product rating output models based on sales history data, posting quantitative data, and keyword data, and then stack the data output from the constructed product rating output models to construct a product prediction model. In other words, by considering various information related to the influencer, more suitable product recommendations can be made for the influencer.

[0172] In addition, the method and device for recommending products and predicting sales volume of influencers according to some embodiments divides learning data into specific period units during the learning process and builds a product rating output model for each, and builds a product prediction model based on the result data, so that the product prediction model can proceed with learning by considering even time-series changes in input data.

[0173] In addition, a method and device for recommending products and predicting sales volume of influencers according to some embodiments can identify other influencers with a high relationship with the influencer, determine their sales volume by considering the sales characteristics of the products, and predict the sales volume of products recommended to the influencer.

[0174]

[0175] FIG. 18 is a diagram illustrating the hardware configuration of a financial server that performs a method for recommending products and predicting sales volume of influencers according to some embodiments of the present invention.

[0176] Referring to FIG. 18, a financial server (100) that performs a method for recommending products and predicting sales volume of influencers according to some embodiments of the present invention may include a processor (101), an input / output device (102), a memory (103), an interface (104), storage (105), and a bus (106). The processor (101), the input / output device (102), the memory (103), the interface (104), and / or the storage (105) may be coupled to each other via a bus (106). The bus (106) corresponds to a path through which data is transferred.

[0177] Specifically, the processor (101) may include at least one of a CPU (Central Processing Unit), an MPU (Micro Processor Unit), an MCU (Micro Controller Unit), a GPU (Graphics Processing Unit), a microprocessor, a digital signal processor, a microcontroller, an application processor (AP), and logic elements capable of performing functions similar thereto.

[0178] The input / output device (102) may include at least one of a keypad, a keyboard, a touch screen, and a display device.

[0179] The memory (103) can load data and / or programs, etc. At this time, the memory (103) is an operating memory for improving the operation of the processor (101), and may include high-speed DRAM and / or SRAM, etc. The memory (103) may include one or more volatile memory devices such as DDR SDRAM (Double Data Rate Static DRAM), SDR SDRAM (Single Data Rate SDRAM), and / or one or more non-volatile memory devices such as EEPROM (Electrically Erasable Programmable ROM), flash memory.

[0180] The interface (104) may perform a function of transmitting data to or receiving data from a communication network. The interface (104) may be wired or wireless. For example, the interface (104) may include an antenna or a wired or wireless transceiver.

[0181] Storage (105) can store and preserve data and / or programs. Storage (105) may include one or more non-volatile memory devices, such as a solid state drive (SSD), a hard drive, or flash memory. In the present invention, storage (105) may store a computer program comprising instructions for performing a method for recommending products for sale and predicting sales volume for influencers.

[0182] In an embodiment, the computer program comprises: a data collecting operation for collecting a plurality of sales history data corresponding to each of a plurality of influencers and a plurality of SNS posting data corresponding to each of the plurality of influencers, wherein the plurality of sales history data includes a sale item, a sale date, and a sale quantity, and the SNS posting data includes a posting date, a posting content, and a feedback content; an operation for analyzing the plurality of SNS posting data to generate a plurality of keyword data and a plurality of posting quantitative data, the operation including extracting advertisement posting data for advertising and selling a product from among the plurality of SNS posting data, identifying an advertised product from the advertisement posting data, and analyzing the advertisement posting data to generate a plurality of keyword data and a plurality of posting quantitative data; an operation for storing the plurality of sales history data, the plurality of keyword data, and the plurality of posting quantitative data for each of a plurality of influencers; an operation for configuring the plurality of sales history data as a first learning dataset to train a first product rating output model, configuring the plurality of keyword data as a second learning dataset to train a second product rating output model, and configuring the plurality of posting quantitative data as a third learning dataset to train a third product rating output model; An operation of stacking the first output data of the learned first product rating output model, the second output data of the learned second product rating output model, and the third output data of the learned third product rating output model to form prediction model learning data; An operation of learning a deep learning-based product prediction model that recommends products for sale to an influencer based on the prediction model learning data, and building the product recommendation model composed of the first product rating output model, the second product rating output model, the third product rating output model, and the product prediction model;An operation of outputting a recommended product for a target influencer through the product recommendation model; and an operation of calculating a predicted sales volume of the target influencer for the outputted recommended product, wherein the operation may include calculating the predicted sales volume of the target influencer for the recommended product by taking into account the relationship between the target influencer and a neighboring influencer and the previous sales volume of the recommended product of the neighboring influencer.

[0183] FIG. 18 is a diagram illustrating a hardware implementation of a device that performs a method for recommending products and predicting sales volume of influencers according to some embodiments of the present invention.

[0184] Referring to FIG. 18, a server (100) according to some embodiments of the present invention may be implemented as an electronic device (1000). The electronic device (1000) may include a controller (1010), an input / output device (1020), a memory device (1030), an interface (1040), and a bus (1050). The controller (1010), the input / output device (1020), the memory device (1030), and / or the interface (1040) may be coupled to each other via a bus (1050). In this case, the bus (1050) corresponds to a path through which data is transferred.

[0185] Specifically, the controller (1010) may include at least one of a CPU (Central Processing Unit), an MPU (Micro Processor Unit), an MCU (Micro Controller Unit), a GPU (Graphics Processing Unit), a microprocessor, a digital signal processor, a microcontroller, an application processor (AP), and logic elements capable of performing functions similar thereto.

[0186] The input / output device (1020) may include at least one of a keypad, a keyboard, a touchscreen, and a display device.

[0187] The memory device (1030) can store data and / or programs, etc.

[0188] The interface (1040) may perform a function of transmitting data to or receiving data from a communication network. The interface (1040) may be wired or wireless. For example, the interface (1040) may include an antenna or a wired / wireless transceiver. Although not illustrated, the memory device (1030) may further include high-speed DRAM and / or SRAM as an operating memory for improving the operation of the controller (1010). The memory device (1030) may store programs or applications therein.

[0189] The server (100) according to embodiments of the present invention may be a system formed by connecting multiple electronic devices (1000) to each other via a network. In this case, each module or combination of modules may be implemented as an electronic device (1000). However, the present embodiment is not limited thereto.

[0190] Additionally, the server (100) may be implemented as at least one of a workstation, a data center, an internet data center (IDC), a direct attached storage (DAS) system, a storage area network (SAN) system, a network attached storage (NAS) system, a redundant array of inexpensive disks (RAID) system, and an electronic document management (EDMS) system, but the present embodiment is not limited thereto.

[0191]

[0192] The method for recommending products and predicting sales volume by an influencer according to an embodiment may also be implemented in the form of a computer-readable medium that stores computer-executable commands and data. In this case, the commands and data may be stored in the form of program code, and when executed by a processor, may generate a predetermined program module to perform a predetermined operation. In addition, the computer-readable medium may be any available medium that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. In addition, the computer-readable medium may be a computer storage medium, and the computer storage medium may include both volatile and nonvolatile, removable and non-removable media implemented using any method or technology for storing information such as computer-readable commands, data structures, program modules, or other data. For example, the computer storage medium may be a magnetic storage medium such as an HDD or SSD, an optical storage medium such as a CD, DVD, or Blu-ray disc, or a memory included in a server accessible via a network.

[0193] Additionally, the influencer's method for recommending products and predicting sales volume according to an embodiment may be implemented as a computer program (or computer program product) containing computer-executable instructions. The computer program includes programmable machine instructions processed by a processor and may be implemented in a high-level programming language, an object-oriented programming language, assembly language, or machine language. Furthermore, the computer program may be recorded on a tangible computer-readable recording medium (e.g., memory, a hard disk, a magnetic / optical medium, or a solid-state drive (SSD).

[0194] Accordingly, the influencer's method for recommending products and predicting sales volume according to an embodiment can be implemented by executing the above-described computer program on a computing device. The computing device may include at least one of a processor, memory, a storage device, a high-speed interface connecting the memory and a high-speed expansion port, and a low-speed interface connecting the low-speed bus and the storage device. Each of these components is interconnected using various buses and may be mounted on a common motherboard or in another suitable manner.

[0195] Here, the processor can process instructions within the computing device, such as instructions stored in a memory or storage device to display graphical information for providing a graphical user interface (GUI) on an external input / output device, such as a display connected to a high-speed interface. In another embodiment, multiple processors and / or multiple buses may be utilized, as appropriate, together with multiple memories and memory types. The processor may also be implemented as a chipset comprising multiple independent analog and / or digital processors.

[0196] Memory also stores information within a computing device. For example, memory may consist of volatile memory units or a collection of volatile memory units. For another example, memory may consist of nonvolatile memory units or a collection of nonvolatile memory units. Memory may also be another form of computer-readable media, such as magnetic or optical disks.

[0197] A storage device can provide a large amount of storage space to a computing device. The storage device can be a computer-readable medium or a configuration including such a medium, and can include, for example, devices within a storage area network (SAN) or other configurations, and can be a floppy disk device, a hard disk device, an optical disk device, a tape device, flash memory, or other similar semiconductor memory device or device array.

[0198] The above description is merely an example of the technical idea of ​​the present embodiment, and those skilled in the art will appreciate that various modifications and variations can be made without departing from the essential characteristics of the present embodiment. Therefore, the present embodiments are not intended to limit the technical idea of ​​the present embodiment, but rather to explain it, and the scope of the technical idea of ​​the present embodiment is not limited by these embodiments. The scope of protection of the present embodiment should be interpreted by the claims below, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of rights of the present embodiment.

Claims

1. A data collection unit that collects a plurality of sales history data corresponding to each of a plurality of influencers and a plurality of SNS posting data corresponding to each of the plurality of influencers, wherein the plurality of sales history data includes a sales item, a sales date, and a sales quantity, and the SNS posting data includes a posting date, posting content, and feedback content; A posting data processing unit that analyzes the plurality of SNS posting data, extracts advertisement posting data for advertising and selling a product from the plurality of SNS posting data, identifies the advertised product from the advertisement posting data, and analyzes the advertisement posting data to generate a plurality of keyword data and a plurality of posting quantitative data; A data storage unit that stores the plurality of sales history data, the plurality of keyword data, and the plurality of posting quantitative data for each of the plurality of influencers; A learning unit that learns a product recommendation model using data stored in the above data storage unit; and Including a sales volume prediction unit that predicts the sales volume of recommended products output from the product recommendation model in response to the target influencer, The above learning department, The above multiple sales history data are configured as a first learning dataset to train a first product rating output model, the above multiple keyword data are configured as a second learning dataset to train a second product rating output model, and the above multiple posting quantitative data are configured as a third learning dataset to train a third product rating output model. The prediction model learning data is configured by stacking the first output data of the learned first product rating output model, the second output data of the learned second product rating output model, and the third output data of the learned third product rating output model. By learning a deep learning-based product prediction model that recommends products for sale to influencers based on the above prediction model learning data, the product recommendation model is constructed, which is composed of the first product rating output model, the second product rating output model, the third product rating output model, and the product prediction model. The above sales volume prediction unit calculates the predicted sales volume of the target influencer for the recommended product by considering the relationship between the target influencer and the neighboring influencer and the previous sales volume for the recommended product of the neighboring influencer. Server.

2. In paragraph 1, The above data collection unit collects SNS posting data by crawling data posted on the SNS accounts of multiple influencers at regular intervals. The above posting content is data crawled from the title and text of the post uploaded by the corresponding influencer, and includes at least one image and video included in the post. The above feedback content is characterized by being information related to the reaction of at least one of the follower, the influencer, and other influencers in response to the post. Server.

3. In paragraph 2, The above multiple keyword data is data that identifies the main keywords of the SNS posting data and counts the frequency of the identified main keywords. The above posting quantitative data is data that quantitatively analyzes the followers' reactions to the corresponding SNS posting data, the relationship with the followers, the number of postings, and the posting format. Server.

4. In paragraph 3, The above learning department, A first influencer-product rating matrix is ​​constructed based on the above sales history data, a first influencer similarity is calculated based on the above sales history data, and a first product rating output model is constructed that outputs an expected rating of the product for each influencer based on the first influencer-product rating matrix and the first influencer similarity. A second influencer-product rating matrix is ​​constructed based on the above keyword data, a second influencer similarity is calculated based on the above sales history data, and a second product rating output model is constructed that outputs an expected rating of the product for each influencer based on the second influencer-product rating matrix and the second influencer similarity. A third influencer-product rating matrix is ​​constructed based on the above posting quantitative data, a third influencer similarity is calculated based on the above sales history data, and a third product rating output model is constructed that outputs the expected rating of the product for each influencer based on the third influencer-product rating matrix and the third influencer similarity. Server.

5. In paragraph 4, The above learning department, The first product rating output model, the second product rating output model, and the third product rating output model are constructed by further considering the average rating of each influencer and the average rating of other influencers, respectively. Server.

6. In paragraph 5, The above learning department, The above prediction model learning data is divided into learning data and verification data, and cross-validation is performed to learn the product prediction model by cross-changing the learning data and the verification data. Server.

7. In paragraph 4, The above sales forecast section, Based on the first influencer similarity, the second influencer similarity, and the third influencer similarity, identifying the neighboring influencer among the plurality of influencers. Server.

8. In paragraph 7, The above sales forecast section, The average market cycle of the above recommended product is calculated, and the base period for considering the previous sales volume of the above recommended product is determined based on the calculated average market cycle. Characterized in that the predicted sales volume of the target influencer for the recommended product is calculated based on the previous sales volume of the recommended product previously sold by the neighboring influencer during the above-determined reference period. Server.

9. A data collecting step of collecting a plurality of sales history data corresponding to each of a plurality of influencers and a plurality of SNS posting data corresponding to each of the plurality of influencers, wherein the plurality of sales history data includes a sales item, a sales date, and a sales quantity, and the SNS posting data includes a posting date, posting content, and feedback content; A step of analyzing the plurality of SNS posting data to generate a plurality of keyword data and a plurality of posting quantitative data, the step including extracting advertisement posting data for advertising and selling a product from the plurality of SNS posting data, identifying the advertised product from the advertisement posting data, and analyzing the advertisement posting data to generate a plurality of keyword data and a plurality of posting quantitative data; A step of storing the plurality of sales history data, the plurality of keyword data, and the plurality of posting quantitative data for each of the plurality of influencers; A step of configuring the plurality of sales history data as a first learning dataset to train a first product rating output model, configuring the plurality of keyword data as a second learning dataset to train a second product rating output model, and configuring the plurality of posting quantitative data as a third learning dataset to train a third product rating output model; A step of configuring prediction model learning data by stacking the first output data of the learned first product rating output model, the second output data of the learned second product rating output model, and the third output data of the learned third product rating output model; A step of learning a deep learning-based product prediction model that recommends products for sale to an influencer based on the above prediction model learning data, and constructing the product recommendation model, which is composed of the first product rating output model, the second product rating output model, the third product rating output model, and the product prediction model; A step of outputting recommended products for the target influencer through the product recommendation model; and A step for calculating the predicted sales volume of the target influencer for the recommended product outputted above, comprising a step for calculating the predicted sales volume of the target influencer for the recommended product by considering the relationship between the target influencer and the neighboring influencer and the previous sales volume of the recommended product of the neighboring influencer. How influencers recommend products and predict sales volume.

10. A computer program stored in a medium for executing a method for recommending products for sale and predicting sales volume by an influencer according to Article 9, in combination with hardware.

Images