KR20230068897A - Method for providing influence analysis service of influencer - Google Patents

Abstract

A method for providing an influencer influence analysis service is provided, including requesting subscriber analysis by entering the influencer's ID in a search terminal, extracting at least one type of reaction data attached to the influencer's posts, Analyzing at least one kind of response data to calculate an influencer's influence measure and transmitting the calculated influence measure to a search terminal.

Description

How to provide influencer influence analysis service {METHOD FOR PROVIDING INFLUENCE ANALYSIS SERVICE OF INFLUENCER}

The present invention relates to a system for providing an influencer influence analysis service, and provides a method of predicting actual influencers among influencers' subscribers, excluding imaginary ones, using reaction data attached to posts.

Social media content continues to grow based on the generalization of digital communication, and the activities of influencers who produce social media content are attracting attention. Organizations such as organizations and companies are executing strategic marketing communications using influencers. Low cost and high efficiency can be obtained, and trust with customers and organizations is proportionally accumulated by maintaining a high relationship with potential customers. because it becomes The ongoing relationship between influencers and users also affects the relationship with brands exposed through content. SNS influencers, who influence not only major customers but also potential customers, are proving effectiveness by conducting specific individual-centered marketing activities. In particular, among social media, influencer marketing activities using YouTube are showing a continuous increase. However, the YouTube channel is evaluated to create marketing effects through the diversity of service and content channels as well as interactive characteristics through interaction.

At this time, a method of calculating the authenticity index of an advertisement using the influencer's influence or user's reaction or evaluating the influencer with social media data has been researched and developed. In this regard, Korea Patent Publication 2021- In No. 0067575 (published on June 08, 2021) and Korean Patent Publication No. 2021-0074080 (published on June 21, 2021), advertisement viewing information posted on social media is received from an influencer terminal, and advertising campaigns The number of interests and comments is extracted, different weights are applied to the number of interests and the number of comments, and the total is calculated as an advertisement response index. and calculates the authenticity index by dividing the advertisement response index by the average response index, generates influencer information including the social media address of the influencer, and collects social media data from the influencer terminal to influence the influence of the influencer. A configuration for evaluating the reliability of the influencer information, receiving advertisement order information from an advertiser terminal, and extracting influencer information corresponding to the advertisement order information is disclosed.

However, in the case of the former, the content of comments or subscribers who expressed interest are not analyzed qualitatively, but simply quantitative analysis, which considers only the number of comments and interest. Information on how many subscribers an influencer actually influences, such as whether it is a valid active user who is actually active on the website, whether it is a false or fake account, is unknown. Even in the latter case, reliability is measured based on the matching rate with preset effective keywords, the number of recommendations, the number of shares, the number of subscribers, and the number of comments, so this is also just a quantitative analysis. Therefore, by qualitatively filtering influencer subscribers with imaginary numbers, the number of subscribers who are actually influenced is calculated, and the right influencer can be selected by exposure to advertisers, and leakage of marketing costs is minimized. Research and development of a platform that can do this is required.

An embodiment of the present invention calculates the average number of reactions, which is the average of the reactions of the influencer's recent posts, analyzes the nationality and language of the users who contributed to the reactions to the recent posts, and extracts the number of subscribers active in Korea, Substitute the extracted nationality and language for the average number of responses to secure the number of subscribers that influence reaches, multiply the value obtained by dividing the average number of subscribers by the average number of responses by the number of users that influence reaches, In case an account is created in a language, it can be derived as an influencer's influence scale by multiplying the weight, and the number of subscribers who are actually influenced by filtering the influencer's subscribers with imaginary numbers through qualitative analysis It is possible to provide an influencer influence analysis service providing system capable of calculating and exposing to advertisers to select suitable influencers and minimizing leakage of marketing costs. However, the technical problem to be achieved by the present embodiment is not limited to the technical problem as described above, and other technical problems may exist.

As a technical means for achieving the above-described technical problem, an embodiment of the present invention is a step of requesting subscriber analysis by inputting an influencer's ID in a search terminal, at least one type attached to the influencer's post The method includes extracting reaction data of, calculating an influencer's influence measure by analyzing at least one type of reaction data, and transmitting the calculated influence measure to a search terminal.

According to any one of the above-described problem solving means of the present invention, the number of subscribers with real influence is selected by filtering the subscribers of influencers with imaginary numbers through qualitative analysis, and suitable influencers are exposed to advertisers. can be selected and the leakage of marketing costs can be minimized.

1 is a diagram for explaining a system for providing an influencer influence analysis service according to an embodiment of the present invention.
FIG. 2 is a block diagram illustrating an analysis service providing server included in the system of FIG. 1 .
3 and 4 are diagrams for explaining an embodiment in which an influencer influence analysis service according to an embodiment of the present invention is implemented.
5 is an operational flowchart illustrating a method for providing an influencer influence analysis service according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail so that those skilled in the art can easily practice the present invention with reference to the accompanying drawings. However, the present invention may be embodied in many different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case where it is "directly connected" but also the case where it is "electrically connected" with another element interposed therebetween. . In addition, when a part "includes" a certain component, this means that it may further include other components, not excluding other components, unless otherwise stated, and one or more other characteristics. However, it should be understood that it does not preclude the possibility of existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

As used throughout the specification, the terms "about", "substantially", etc., are used at or approximating that value when manufacturing and material tolerances inherent in the stated meaning are given, and do not convey an understanding of the present invention. Accurate or absolute figures are used to help prevent exploitation by unscrupulous infringers of the disclosed disclosure. The term "step of (doing)" or "step of" as used throughout the specification of the present invention does not mean "step for".

In this specification, a "unit" includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Further, one unit may be realized using two or more hardware, and two or more units may be realized by one hardware. On the other hand, '~ unit' is not limited to software or hardware, and '~ unit' may be configured to be in an addressable storage medium or configured to reproduce one or more processors. Thus, as an example, '~unit' refers to components such as software components, object-oriented software components, class components, and task components, processes, functions, properties, and procedures. , subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays and variables. Functions provided within components and '~units' may be combined into smaller numbers of components and '~units' or further separated into additional components and '~units'. In addition, components and '~units' may be implemented to play one or more CPUs in a device or a secure multimedia card.

In this specification, some of the operations or functions described as being performed by a terminal, device, or device may be performed instead by a server connected to the terminal, device, or device. Likewise, some of the operations or functions described as being performed by the server may also be performed by a terminal, apparatus, or device connected to the server.

In this specification, some of the operations or functions described as mapping or matching with the terminal mean mapping or matching the terminal's unique number or personal identification information, which is the terminal's identifying data. can be interpreted as

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram for explaining a system for providing an influencer influence analysis service according to an embodiment of the present invention. Referring to FIG. 1 , an influencer influence analysis service providing system 1 may include at least one search terminal 100, an analysis service providing server 300, and at least one social media platform 400. . However, since the influencer influence analysis service providing system 1 of FIG. 1 is only one embodiment of the present invention, the present invention is not limitedly interpreted through FIG. 1 .

At this time, each component of FIG. 1 is generally connected through a network (Network, 200). For example, as shown in FIG. 1 , at least one search terminal 100 may be connected to an analysis service providing server 300 through a network 200 . In addition, the analysis service providing server 300 may be connected to at least one search terminal 100 and at least one social media platform 400 through the network 200 . In addition, at least one social media platform 400 may be connected to the analysis service providing server 300 through the network 200 .

Here, the network refers to a connection structure capable of exchanging information between nodes such as a plurality of terminals and servers, and examples of such networks include a local area network (LAN) and a wide area network (WAN: Wide Area Network), the Internet (WWW: World Wide Web), wired and wireless data communications networks, telephone networks, and wired and wireless television communications networks. Examples of wireless data communication networks include 3G, 4G, 5G, 3rd Generation Partnership Project (3GPP), 5th Generation Partnership Project (5GPP), Long Term Evolution (LTE), World Interoperability for Microwave Access (WIMAX), Wi-Fi , Internet (Internet), LAN (Local Area Network), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network), PAN (Personal Area Network), RF (Radio Frequency), Bluetooth (Bluetooth) network, NFC ( A Near-Field Communication (Near-Field Communication) network, a satellite broadcasting network, an analog broadcasting network, a Digital Multimedia Broadcasting (DMB) network, etc. are included, but not limited thereto.

In the following, the term at least one is defined as a term including singular and plural, and even if at least one term does not exist, each component may exist in singular or plural, and may mean singular or plural. It will be self-evident. In addition, the singular or plural number of each component may be changed according to embodiments.

At least one search terminal 100 inputs the ID or account of the influencer using a web page, app page, program or application related to the influencer influence analysis service, and then the imaginary number of subscribers of the influencer , that is, it may be a terminal of an individual, company, institution, etc. who wants to check the influence scale of how many subscribers are actually influencing the rest of the subscribers excluding fake or fake accounts. At this time, the searcher of the search terminal 100 may be a person who searches for an influencer for fun, or a person who has an interest in analyzing subscribers and understanding a measure of influence in order to request an advertisement.

Here, at least one search terminal 100 may be implemented as a computer capable of accessing a remote server or terminal through a network. Here, the computer may include, for example, a laptop, a desktop, a laptop, and the like equipped with a navigation system and a web browser. In this case, at least one search terminal 100 may be implemented as a terminal capable of accessing a remote server or terminal through a network. At least one search terminal 100 is, for example, a wireless communication device that ensures portability and mobility, and includes navigation, PCS (Personal Communication System), GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet ) may include all types of handheld-based wireless communication devices such as terminals, smartphones, smart pads, tablet PCs, and the like.

The analysis service providing server 300 may be a server that provides an influencer influence analysis service web page, app page, program or application. In addition, the analysis service providing server 300 stores mathematical formulas and algorithms capable of calculating the influence scale, and when an influencer's ID or account name is entered in the search terminal 100, the social media platform 400 ) and search the influencer's page. In addition, the analysis service providing server 300 selects N recent posts in the influencer's page, for example, the Post page in the case of Facebook, and the reaction data attached to the recent posts, that is, likes, comments, shares, etc. may be a server that collects In addition, the analysis service providing server 300 may be a server that determines the number of Korean subscribers after determining the nationality and language of the subscribers who have responded. Recently, influencer accounts with false or fraudulent subscribers have been traded with money, and accounts with foreign subscribers are traded cheaper than accounts with Korean subscribers. In Korea, the user authentication process using a mobile phone is common, but in other countries, as long as there is an e-mail address, one person can create false subscriber accounts and attach them. Therefore, the analysis service providing server 300 may be a server that outputs an influence measure to the search terminal 100 based on the average number of responses, the average number of subscribers, and parameters of domestic subscribers after determining the number of Korean subscribers.

Here, the analysis service providing server 300 may be implemented as a computer capable of accessing a remote server or terminal through a network. Here, the computer may include, for example, a laptop, a desktop, a laptop, and the like equipped with a navigation system and a web browser.

At least one social media platform 400 uses the influencer influence analysis service-related web page, app page, program, or application, or the influencer's ID or account name in the analysis service providing server 300. If you search, it could be a social media platform that serves the resulting influencer page. At this time, the social media may be Facebook, Instagram, Twitter, TikTok, YouTube, etc., but is not limited thereto. Here, at least one social media platform 400 may be implemented as a computer capable of accessing a remote server or terminal through a network. Here, the computer may include, for example, a laptop, a desktop, a laptop, and the like equipped with a navigation system and a web browser.

2 is a block diagram illustrating an analysis service providing server included in the system of FIG. 1, and FIGS. 3 and 4 show an embodiment in which an influencer influence analysis service is implemented according to an embodiment of the present invention. It is a drawing for explanation.

Referring to FIG. 2, the analysis service providing server 300 includes a request unit 310, an extraction unit 320, a calculation unit 330, a transmission unit 340, a product management unit 350, and an imaginary number determination unit 360. ) may be included.

The analysis service providing server 300 according to an embodiment of the present invention or another server (not shown) operating in conjunction with at least one search terminal 100 and at least one social media platform 400 influence influencers. When an analysis service application, program, app page, web page, etc. is transmitted, at least one search terminal 100 and at least one social media platform 400, an influencer influence analysis service application, program, app page, You can install or open web pages, etc. In addition, a service program may be driven in at least one search terminal 100 and at least one social media platform 400 by using a script executed in a web browser. Here, the web browser is a program that allows users to use the web (WWW: World Wide Web) service, and means a program that receives and displays hypertext described in HTML (Hyper Text Mark-up Language). For example, Netscape , Explorer, Chrome, and the like. In addition, an application means an application on a terminal, and includes, for example, an app running on a mobile terminal (smart phone).

Referring to FIG. 2, prior to describing FIG. 2, a background from which an embodiment of the present invention was derived will be described.

Recently, companies that increase the number of followers (Followers, Subscribers, Subscribers) on social media are prevalent. Most places advertise that it takes 5 hours to increase 250 followers and 40,000 won. There are 250 Korean followers (39,800 won), and Instagram followers are about 160 won per person. Social media marketing companies are advertising that they will definitely increase the number of followers within 24 hours, follow more than promised in preparation for unfollowing (act of unfollowing), and if the number decreases by more than 5% within a month, A/S It is refilled with The followers filled by the company are all pretending to be Koreans as described in the product description. However, it was an account that randomly spliced photos and writings of other people's Instagram users, but looking at the entire follower list, you can't tell the difference between real and fake.

There is no place where custom manipulation is not possible. The same goes for Facebook. Even if you run a Facebook page introducing restaurants and want to quickly increase your followers, you can get a price tag of 70 won per Korean follower, 30 won for ghost Koreans, and 20 won for foreigners. The manipulator company advises to recommend Korean ghost followers rather than Southeast Asian foreign followers, although the price is cheap, but it is suspicious. Competition between companies is fierce. With the know-how accumulated over the years, it avoids surveillance on Facebook and Instagram and uploads with the number of 'Likes' and comments. Not only Gram and YouTube, but also Tumblr, Sound Cloud, and Vimeo, all of which can be manipulated, and each company has a different competitive edge.

The online music platform SoundCloud also manipulates the number of plays, and the price is 5,000 won per 10,000 times. In addition, Click Farm, which was caught in Thailand, was operated with 500 smartphones and 400,000 SIM cards, and manipulated a large amount of likes and shares. Three Chinese people who came to Thailand repeatedly shared and liked the advertisement posts to promote a specific Chinese product here. In addition, Mediakicks, an American marketing company, experimented with how far fakes made by manipulation would work. It was an experiment in which they created two accounts of a fashion model and a traveler on Instagram and constantly purchased fake followers. When the number of followers exceeded 10,000, advertisement offers came in, and advertisers advertised themselves as influencers. The fake account won two advertisements each for swimwear and food and beverage in three months. Hong Moon-ki, general affairs director of the Korea Public Relations Society, and professor at Han Sei-dae, said that business models that profit from fake followers undermine the basic order of the market, and eventually the damage will go to consumers. He said that trust is being lost as this spreads, and that a system that more actively identifies active users and fakes is needed.

Accordingly, in one embodiment of the present invention, rather than quantitative data such as the number of subscribers, the number of shares, the number of comments, and the number of likes, the number of subscribers that the actual influencer has influence is qualitatively analyzed, so that individuals or stakeholders to provide to

Referring to FIG. 2 , the requesting unit 310 may request subscriber analysis by inputting the ID of the influencer in the search terminal 100 . For example, if there is an influencer called "Tehran, the center of law" and the social media is Instagram, @teheran_law is the influencer's ID or account.

The extraction unit 320 may extract at least one type of reaction data attached to the post of the influencer. At this time, the response data may be the number of likes, but may also be the number of shares, the number of comments, and the like.

The calculation unit 330 may analyze at least one type of response data to calculate an influencer's influence scale. When the calculation unit 330 analyzes at least one type of response data to calculate an influencer's influence scale, it may extract a predetermined number of posts by arranging the posts of the influencer in the most recent order. For example, if the preset number of posts is 10, 10 posts may be extracted by arranging in the most recent order.

The calculation unit 330 may calculate an average number of responses, which is an average of at least one response data included in a predetermined number of extracted posts. At this time, the response of at least one response data may be like. For example, there are 10 posts, the number of reaction data for the 1st post, e.g. the "Like" button was clicked A times, the 2nd post is B times,...,the 10th post is clicked J times. , the average number of responses is the result of (A+B+....+J) divided by 10. The calculation unit 330 may extract the number of domestic subscribers based on the nationality and language of subscribers to whom at least one response data attached to a predetermined number of posts is attached. Domestic subscribers may preferably be Korean subscribers. For example, in order to download Facebook, Instagram, TikTok, YouTube, etc., you basically need an account (e-mail address, etc.) to sign up for Google Play on Android and Apple Store on iOS. Then, in order to sign up for Facebook, an e-mail address and password are required again, and through these two, the nationality of the registered user can be identified.

In addition, the number of domestic subscribers is a number extracted with an upper limit of 500, and only the nationality and language of up to 500 subscribers (not domestic subscribers) are extracted to speed up the data acquisition speed. Also, before multiplying the number of domestic subscribers by the average number of responses, the number of domestic subscribers can be divided by 500. The calculation unit 330 may multiply the average number of responses by the number of domestic subscribers, then multiply by a value obtained by dividing the average number of subscribers of influencers by the average number of responses, and then multiply by a preset weight. The preset weight is a weight added in case a domestic subscriber creates an account in a language other than Korean, and the preset weight may be 1.7. Recently, there are applications that have not yet entered the corresponding country due to patent rights, copyright issues, or business rights not being obtained in the corresponding country. Accordingly, since each country supports different applications, each country can be downloaded whenever it changes. The application is open or restricted.

Accordingly, in recent years, even Koreans often change their accounts to US or Australian accounts in order to download applications that cannot be used in Korea, for example, music play applications such as Spotifiy. Therefore, a weight is multiplied in case a Korean uses a foreign account. For example, a Korean, Mr. A, signed up for Google Play using an American account, downloaded Facebook, and clicked "Like" while using Facebook. not classified as In this case, it is excluded. In order to prepare for this, the number of actual domestic subscribers is multiplied by a weight of 1.7 times. That is, if 100 people are filtered as domestic subscribers, about 70 people are mistaken for foreign (overseas) users, and they are filtered and filtered, and they are filled back. Summarizing this as an equation, the following equation 1 is obtained.

In this case, the average number of subscribers may mean the average number of subscribers during a preset period or may mean the total number of subscribers based on the present. The weight is 1.7 as described above, but may be changed according to embodiments. Here, domestic accounts may be extracted by analyzing the location tag of the account and the language of the post in addition to or replacing the multiplication of the weight . For example, if the account is "Denworth" in English, the language of the post is in Korean with the text "The weather is nice", and the hashtags on the post are #Itaewon, #Songnidangil, this is considered a Korean account and the number of domestic subscribers may also include

The transmitter 340 may transmit the calculated influence scale to the search terminal 100 . The influence scale may be calculated as a numerical value and transmitted to the search terminal 100 . At this time, the search result may be reset and the search may be performed again, but visualization may also be possible in which search results accumulated by the searcher as a history log and subsequent search results are compared and analyzed. For example, if you searched for influencers A to Z, you could sort them in descending order on an influencer scale, and if the search results are not always the same (because you analyze 10 recent posts), they fluctuate periodically or in real time. You can also rearrange the rankings by reflecting the data. In addition, changes in the influence scale of the influencer you searched can be provided as a chart, such as a stock graph. After a search by crawling, information can be provided by tagging a keyword at the point in time when the graph dropped for the reason.

The product management unit 350 analyzes at least one type of response data in the calculation unit 330 to calculate the influence scale of the influencer, and then selects a predetermined number of categories of posts using a pre-stored image analysis algorithm. can be extracted and classified. In this case, the image analysis algorithm may use deep learning for recognizing and extracting objects in the image.

<Deep Learning>

Recently, as a method of object recognition in an image, a method using artificial intelligence is frequently used, and the most widely used method among them is an R-CNN series algorithm. Several neural networks are continuously connected in a method using region-based convolution, and there are weight values corresponding to each neural network. This weight value is determined by learning, and the values determined in this way are used for object recognition. Fast R-CNN, Faster R-CNN, and Mask R-CNN exist.

The first Fast R-CNN can secure a Backbone Network through single-step learning. Although it has the advantage of being able to obtain an artificial intelligence model simply because it can perform single-step learning, it has the disadvantage that it takes time to generate a large number of Region of Interest (ROI) Boundary Box candidates. This is mainly due to the performance of a selective search algorithm with a long computation time.

In order to improve this point, Faster R-CNN has proposed a new method, in particular, a new deep neural network is used to generate a candidate group of bounding boxes for the region of interest. That is, conventional algorithms mainly use CPU resources and thus require a relatively large amount of computation time, but deep neural networks reduce computation time because GPUs are used. Such a deep neural network is called a region proposal network because it functions to provide a candidate group for a region of interest, and has relatively few neural network layers compared to the backbone network. Most of the hardware resources required for object recognition can be moved from the CPU to the GPU through this neural network. However, in order to create a complete model with such a structure, it is necessary to perform learning through several stages. Faster R-CNN consists of a backbone network, a region proposal network, and a classifier network, and since several steps are required to learn these three deep neural networks, it takes time to obtain a completed model with a multi-step learning technique. may be extended.

The biggest feature of Mask R-CNN, which is the last of the R-CNN series methods, is that it enables segmentation of object recognition. Previously, object recognition estimated the position of an object through a square box in an image and classified its type, but in Mask R-CNN, object recognition technology designs a deep neural network to distinguish the location of an object in pixel units. The role of this deep neural network is to distinguish whether each pixel corresponds to an object or not, which is called a binary mask. A binary mask is applied to the feature map obtained from the backbone network to classify in pixel units. However, the size of a general feature map is small, ranging from 1/4 to 1/16 the size of the original image. Even if the object area is determined in units of pixels in the reduced feature map, the error appears large in the original image. In order to reduce such an error, a technique called ROI Align is used, which is a method of reducing the decimal point error generated in the ROI pool area by a quadratic linear interpolation method.

Among YOLO, YOLO-v3 is an algorithm in which one convolutional neural network simultaneously predicts several bounding boxes and predicts class probabilities for each bounding box. At this time, the convolutional neural network used is used for generating a feature map, and it is called a backbone network because it is in charge of the backbone of the algorithm. This feature map is composed of several grid cells, and it is an algorithm that simultaneously determines the type and location of a target object by applying a score method to each cell. This algorithm may be configured to recognize the learned object every image frame.

<Generative Adversarial Network (GAN)>

If the big data training dataset and test dataset are not sufficient, the amount of data to be trained can be increased using GAN. For example, StarGAN is a network in which two modules with conflicting purposes compete with each other to generate images, and it is composed of a generator and a discriminator. The generator creates a fake image, and the discriminator is a module that distinguishes the generated image from the real image. The purpose of the generator is to generate a fake image that is realistic enough that the discriminator cannot distinguish it, and the purpose of the discriminator is to distinguish the image generated by the generator from the real image. The generator learned in this way can generate a fake image that is realistic to the extent that the discriminator cannot distinguish whether it is a real image or a generated image. Here, when training the network, it can be learned using the loss value (Loss) of Wasserstein GAN.

The product management unit 350 extracts product names included in a predetermined number of posts, determines category fit, which is the fit between categories and product names, and calculates an influence measure for each product by summing the influence measure and the category fit. For example, let's assume that influencer A is a veterinarian, and the topic of companion animal nutrition has been the mainstream in A's recent posts (posts). At this time, when extracting the category, it can be extracted as companion animal nutrition, and each product name can be extracted as B, C, D, E.., etc. Here, the degree of compatibility between the companion animal nutritional supplement (category) and the product name B, C, D, E... can be identified. If you look at a post on a fake or fake account, there are many cases where MacroBot takes other posts and puts them together, so the category of the post and its content often do not match. Or, there are cases where a true influencer, not a real fake account, posts off-topic because there are many cases where the writing goes in a different direction. In this case, the fit is reduced.

There are times when you need to analyze text in addition to photos (images) in order to understand the category of posts and the suitability of products. To this end, text analysis may be further performed in the present invention. For example, along with data crawling and numerical data analysis, which are quantitative analyses, qualitative text analysis, which is comment or posting analysis, can be performed together. At this time, using the hidden data analysis technique, the goal may be to provide an influencer's influence scale for each product. To this end, after analyzing the accounts of influencers whose posts and comments have an average high correlation, the results are verified with TF-IDF (Term Frequency-Inverse Document Frequency), so that the higher the match, the higher the weight. there is. This means that it is even better if it is used when recommending an influencer to other subscribers as subscribers not only show high interest in the post, but also match the unique characteristics and attributes of the post. In particular, the influence scale for each product can be finally extracted by performing LDA (Latent Dirichlet Allocation)-based similarity-based topic modeling on the TF-IDF (Term Frequency - Inverse Document Frequency) analysis result for each influencer account.

To this end, data collection and visualization can be performed in R language specialized for the corresponding function, and in the case of TF-IDF and LDA analysis, additional Python work can be performed in parallel for the convenience of analysis. The test is largely divided into two steps. The first is the analysis of the relationship between variables through numerical data provided by social media. In the case of PCA, visualization can be performed through scree and loading pictures. Next, a correlation coefficient is obtained for each variable (numerical data) and a meaningful conclusion is drawn through the numerical data.

In the second case, key words for each influencer account are extracted through TF-IDF analysis, LDA topic modeling is performed based on the words, and the influence scale for each product is calculated based on this. Since there is no rating information on Instagram, Facebook, YouTube, etc., in the case of the performance evaluation, those who wrote comments with a high average correlation with the content of the posts for each subject of the influencer account received a high rating. , and vice versa, it is assumed that a lower rating is given and a rating matrix is created. For example, if subscriber A comments on a post registered with influencer B's account that is highly related to the content of the post, the normalization value of the average value is assumed to be the rating given by subscriber A to influencer B's account. .

On Instagram, Facebook, and YouTube, there is a stronger tendency to like more often and not to comment most of the time, but in one embodiment of the present invention, only those posts with comments above the average threshold among high-profile influencer posts were considered. Therefore, it is assumed that there is no problematic sparse matrix problem. If subscriber A commented on several posts from influencer B's account, the average normalized value is used as the rating. When proceeding with the method according to an embodiment of the present invention, beyond the step of simply recommending well-known influencers, among posts with influencer attributes, how much influence influencers have by category and product has the advantage of being able to identify In addition, it is possible to maximize the satisfaction of the marketing effect by recommending highly recognized influencer attributes for each product to companies that want to do influencer-based marketing for new items.

The imaginary number discrimination unit 360 requests an analysis of subscribers by inputting the ID of the influencer from the search terminal 100 in the requesting unit 310, and then determines the number of subscribers, the number of posts, and the engagement rate of the influencer. Based on this, it is possible to check whether the account has a fake or fake follower. At this time, confirmation may be determined by determining whether the ratio of the participation rate to the number of postings exceeds a predetermined allowable range. For example, if there are 10 posts but 1 million likes, it can be determined that this is a fake account. Alternatively, verification can be determined by detecting fake subscribers with at least one type of artificial intelligence algorithm. Recently, methods using artificial intelligence to filter out macrobots or detect fake or fraudulent accounts are increasing. Therefore, before calculating the influence scale, it may be informed in advance what percentage of subscribers are real accounts and what percentage are non-real accounts.

Hereinafter, an operation process according to the configuration of the above-described analysis service providing server of FIG. 2 will be described in detail with reference to FIGS. 3 and 4 as examples. However, it will be apparent that the embodiment is only any one of various embodiments of the present invention, and is not limited thereto.

Referring to FIG. 3, (a) the analysis service providing server 300 receives the influencer's ID from the search terminal 100 and checks the influence scale, as shown in (b), the average response to the last 10 posts. count the number (c) Then, domestic (Korean) subscribers are extracted by extracting language and nationality from responses to 10 posts, and at this time, only up to 500 subscribers are extracted to secure data processing speed. And, when multiplying the number of domestic subscribers and the average number of responses as shown in (d), divide by 500 since only 500 were extracted.

Then, as shown in (a) of FIG. 4, the average number of subscribers is divided by the average number of responses. If this number is 10, it means that 10% of the subscribers responded to the post. Then, as in (b), the number of Koreans included in the 500 is multiplied by 1.7 in preparation for the case that Koreans who created an account with a foreign account and joined social media are eliminated (filtered) in (c) of FIG. It compensates (supplements) the number of Koreans who may have been filtered out by And (c) In order to calculate the influence scale for each product, the category fit can be further calculated and added to the influence scale calculated in (b), and the influence scale value by determining the imaginary number among the influencer's subscribers as shown in (d) may be provided along with

Matters not described for the method of providing the influencer influence analysis service of FIGS. 2 to 4 are the same as those described for the method of providing the influencer influence analysis service through FIG. Since it can be inferred, the following description will be omitted.

5 is a diagram illustrating a process of transmitting and receiving data between components included in the influencer influence analysis service providing system of FIG. 1 according to an embodiment of the present invention. Hereinafter, an example of a process of transmitting and receiving data between each component will be described through FIG. 5, but the present application is not limited to such an embodiment, and according to various embodiments described above, It is obvious to those skilled in the art that a process of transmitting and receiving data may be changed.

Referring to FIG. 5, the analysis service providing server requests subscriber analysis by inputting the influencer's ID in the search terminal (S5100), and extracts at least one type of reaction data attached to the influencer's post. (S5200).

Then, the analysis service providing server analyzes at least one kind of response data to calculate an influencer's influence measure (S5300), and transmits the calculated influence measure to the search terminal (S5400).

The order between the above-described steps (S5100 to S5400) is only an example, and is not limited thereto. That is, the order of the above-described steps (S5100 to S5400) may be mutually changed, and some of the steps may be simultaneously executed or deleted.

Matters that have not been explained about the method of providing the influencer influence analysis service of FIG. 5 are the same as those described for the method of providing the influencer influence analysis service through FIGS. 1 to 4 above, or can be easily obtained from the description Since it can be inferred, the following description will be omitted.

The method for providing an influencer influence analysis service according to an embodiment described with reference to FIG. 5 may be implemented in the form of a recording medium including instructions executable by a computer such as an application or program module executed by a computer. there is. Computer readable media can be any available media that can be accessed by a computer and includes both volatile and nonvolatile media, removable and non-removable media. Also, computer readable media may include all computer storage media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data.

The above-described method for providing an influencer influence analysis service according to an embodiment of the present invention is executed by an application basically installed in a terminal (this may include a program included in a platform or operating system basically installed in the terminal). It may be executed by an application (ie, a program) directly installed in the master terminal by a user through an application providing server such as an application store server, a web server related to an application or a corresponding service. In this sense, the above-described method for providing an influencer influence analysis service according to an embodiment of the present invention is implemented as an application (i.e., a program) that is basically installed in a terminal or directly installed by a user, and is readable by a computer such as a terminal. can be recorded on a recordable medium.

The above description of the present invention is for illustrative purposes, and those skilled in the art can understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, the embodiments described above should be understood as illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present invention is indicated by the following claims rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and equivalent concepts should be construed as being included in the scope of the present invention. do.

Claims (8)

In the analysis service providing method executed in the analysis service providing server,
Requesting subscriber analysis by entering the ID of the influencer in the search terminal;
extracting at least one kind of reaction data attached to the post of the influencer;
calculating an influence scale of the influencer by analyzing the at least one type of response data; and
transmitting the calculated influence measure to the search terminal;
Influencer influence analysis service providing system comprising a.
According to claim 1,
The step of calculating the influence scale of the influencer by analyzing the at least one kind of response data,
extracting a predetermined number of posts by arranging the posts of the influencer in the order of most recent;
Calculating an average number of responses, which is an average of at least one response data included in the predetermined number of extracted posts;
extracting the number of domestic subscribers based on the nationality and language of subscribers to whom at least one response data attached to the predetermined number of posts is attached;
multiplying the number of domestic subscribers by the average number of responses, and then multiplying the average number of subscribers of the influencer by the average number of responses;
Influencer influence analysis service providing system comprising a.
According to claim 2,
The preset number of the preset number of posts is 10,
Influencer influence analysis service providing system, characterized in that the reaction of the at least one reaction data is like.
According to claim 2,
The number of domestic subscribers is a number extracted with an upper limit of 500,
An influencer influence analysis service providing system, characterized in that dividing the number of domestic subscribers by 500 before multiplying the average number of responses by the number of domestic subscribers.
According to claim 1,
After analyzing the at least one kind of response data to calculate the influence scale of the influencer,
extracting and classifying categories of the predetermined number of postings using a pre-stored image analysis algorithm;
extracting product names included in the predetermined number of postings;
determining category suitability, which is the suitability between the category and the product name;
calculating an influence measure for each product by summing the influence measure and the degree of category conformity;
Influencer influence analysis service providing system further comprising a.
According to claim 1,
After the step of requesting subscriber analysis by inputting the influencer's ID in the search terminal,
Checking whether the account has fake or fake followers based on the number of subscribers, the number of posts, and the engagement rate of the influencer;
Influencer influence analysis service providing system further comprising a.
According to claim 6,
The confirmation is an influencer influence analysis service providing system, characterized in that it is determined whether the ratio of the participation rate to the number of postings exceeds a preset allowable range.
According to claim 6,
Influencer influence analysis service providing system, characterized in that the confirmation is determined by a method of detecting fake subscribers with at least one type of artificial intelligence algorithm.

Images