Method, device and system for matching between brand, artist and content based on artificial intelligence

Abstract

According to one embodiment, a method for matching between an artificial intelligence-based brand, an artist, and content, performed by a device comprises: receiving a request for matching an artist and content for a first brand from a first user terminal; obtaining first brand information, which is information related to the first brand; encoding the first brand information to generate a first input signal; inputting the first input signal to a first artificial intelligence model trained to select a recommended artist through the analysis of brand information; when the recommended artist of the first brand is selected as the first artist through the first input signal, obtaining a first output signal representing the first artist from the first artificial intelligence model; determining the recommended artist of the first brand as the first artist based on the first output signal; inputting the first input signal to a second artificial intelligence model trained to select a recommended content type through the analysis of brand information; and when the recommended content type of the first brand is selected as the first type through the first input signal, obtaining a second output signal representing the first type from the second artificial intelligence model. An artificial intelligence-based method for matching between a brand, an artist, and content is provided, comprising: a step of determining the recommended content type of the first brand as the first type based on the second output signal; a step of matching the first brand, the first artist, and the content of the first type to generate first matching information; and a step of transmitting the first matching information to the first user terminal.

Description

Technology Field

[0001] The following embodiments relate to technology for matching brands, artists, and content based on artificial intelligence, and specifically to technology for generating matching information between brand, artist, and content types using an artificial intelligence model. Background Technology

[0002] With the recent growth of the digital advertising market, branded content and PPL (Product Placement)-based marketing are actively being pursued to maximize advertising effectiveness through collaboration between brands and content. In particular, the method of producing content utilizing artists or influencers suitable for a brand's image and target audience, and disseminating it through various online platforms, is being utilized as a key strategy in the advertising and marketing fields.

[0003] However, in conventional advertising collaboration processes, reliance on the experience or intuition of advertising planners or marketing managers was frequently used to determine the suitability between a specific brand and an artist or content type. This could lead to collaboration partners being selected without sufficient consideration of various factors, such as brand image, target audience, and past advertising performance data. Consequently, this approach faced limitations, including significant variability in advertising performance and a higher risk of campaign failure.

[0004] Furthermore, the process of planning collaborations between brands and artists required a comprehensive analysis of various factors, such as brand characteristics, content formats, the artist's image and activity history, and past advertising performance data, which resulted in significant time and cost. In particular, as the scale of advertising campaigns grows and the scope of collaborations diversifies, this analysis process becomes more complex, potentially making it difficult for advertisers to quickly identify the optimal combination of collaborations.

[0005] Meanwhile, although technologies utilizing artificial intelligence to analyze various data and derive recommendation results have recently been advancing, there has not been a sufficient number of technologies proposed that comprehensively analyze brand information, artist characteristics, content types, and past advertising performance data to simultaneously recommend artists and content types suitable for a brand and automatically generate matching information for advertising collaboration based on this.

[0006] Therefore, there is a need to develop technology that uses artificial intelligence based on brand characteristics and related data to analyze the suitability between brands, artists, and content types, and thereby automatically selects artists and content types suitable for the brand to generate matching information. Prior art literature

[0007] Korean Registered Patent No. 10-2643697, Korean Registered Patent No. 10-2515439, Korean Registered Patent No. 10-2414937, Korean Registered Patent No. 10-2338619 The problem to be solved

[0008] According to one embodiment, the purpose is to provide a method, device, and system for matching between an AI-based brand, an artist, and content.

[0009] The objectives of the present invention are not limited to those mentioned above, and other unmentioned objectives will be clearly understood from the description below. means of solving the problem

[0010] According to one embodiment, a method for matching between an artificial intelligence-based brand, an artist, and content, performed by a device comprises: receiving a request for matching an artist and content for a first brand from a first user terminal; obtaining first brand information, which is information related to the first brand; encoding the first brand information to generate a first input signal; inputting the first input signal to a first artificial intelligence model trained to select a recommended artist through the analysis of brand information; when the recommended artist of the first brand is selected as the first artist through the first input signal, obtaining a first output signal representing the first artist from the first artificial intelligence model; determining the recommended artist of the first brand as the first artist based on the first output signal; inputting the first input signal to a second artificial intelligence model trained to select a recommended content type through the analysis of brand information; and when the recommended content type of the first brand is selected as the first type through the first input signal, obtaining a second output signal representing the first type from the second artificial intelligence model. An artificial intelligence-based method for matching between a brand, an artist, and content is provided, comprising: a step of determining the recommended content type of the first brand as the first type based on the second output signal; a step of matching the first brand, the first artist, and the content of the first type to generate first matching information; and a step of transmitting the first matching information to the first user terminal.

[0011] The first artificial intelligence model, based on the first input signal, identifies the first brand information, and based on the first brand information, identifies the brand image, target customer base, and brand campaign performance of the first brand, selects a first artist, which is one of a plurality of pre-registered artists, as the subject of analysis, obtains first artist information, which is information related to the first artist, and based on the first artist information, identifies the artist image, fan base, and artist advertising performance of the first artist, calculates the similarity between the brand image of the first brand and the artist image of the first artist as a first similarity, calculates the similarity between the target customer base of the first brand and the fan base of the first artist as a second similarity, calculates the similarity between the brand campaign performance of the first brand and the artist advertising performance of the first artist as a third similarity, sums the first similarity, the second similarity, and the third similarity to calculate a matching score between the first brand and the first artist, and when the matching scores between the first brand and the plurality of artists are each calculated, compares the matching scores between the first brand and the plurality of artists. As a result, if it is confirmed that the matching score between the first brand and the first artist is the highest, the recommended artist of the first brand can be selected as the first artist.

[0012] The second artificial intelligence model described above verifies the first brand information based on the first input signal, verifies the brand image, target audience, and brand campaign performance of the first brand based on the first brand information, selects a first type as an analysis target which is one of a plurality of pre-registered content types, obtains first content type information which is information related to the first type, verifies the content image, main audience of the content, and content advertising performance of the first type based on the first content type information, calculates the similarity between the brand image of the first brand and the content image of the first type as a fourth similarity, calculates the similarity between the target audience of the first brand and the main audience of the content of the first type as a fifth similarity, calculates the similarity between the brand campaign performance of the first brand and the advertising performance of the content of the first type as a sixth similarity, sums the fourth similarity, the fifth similarity, and the sixth similarity to calculate a matching score between the first brand and the first type, and when the matching scores between the first brand and the plurality of content types are each calculated, the matching between the first brand and the plurality of content types If, as a result of comparing the scores, it is confirmed that the matching score between the first brand and the first type is the highest, the recommended content type of the first brand can be selected as the first type. Effects of the invention

[0013] According to one embodiment, by utilizing an artificial intelligence model trained to analyze brand information, artist and content types suitable for a brand can be automatically recommended, and matching information between the brand, artist, and content can be generated based on this. This reduces the limitations of subjective judgment that may occur in the conventional method of selecting collaboration targets based on the experience or intuition of advertising planners or marketing managers, and enables the derivation of collaboration combinations suitable for a brand based on data.

[0014] In addition, according to one embodiment, recommendation results are derived through a first artificial intelligence model that performs artist recommendation using brand information as input and a second artificial intelligence model that performs content type recommendation, and by combining these, matching information between brands, artists, and content types can be generated, thereby enabling the optimal collaboration combination to be derived more quickly and efficiently during the advertising campaign planning process.

[0015] In addition, according to one embodiment, since the analysis is performed based on various information such as the brand image, target customer base, and past advertising performance data, the suitability between the brand, artist, and content type can be analyzed more precisely, thereby improving the effectiveness of advertising content and increasing the likelihood of success for advertising campaigns.

[0016] In addition, according to one embodiment, matching information between brands, artists, and content can be automatically generated and provided to a user terminal, thereby reducing the time and cost required for advertisers or marketing managers to search for and select collaboration targets and improving the efficiency of advertising planning work.

[0017] Meanwhile, the effects according to the embodiments are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below. Brief explanation of the drawing

[0018] FIG. 1 is a schematic diagram showing the configuration of a system according to one embodiment. FIG. 2 is a flowchart illustrating the matching process between an AI-based brand, an artist, and content according to one embodiment. FIG. 3 is a flowchart for explaining the process of selecting a recommended artist according to one embodiment. FIG. 4 is a flowchart for explaining the process of selecting a recommended content type according to one embodiment. FIGS. 5 and 6 are flowcharts for explaining the process of selecting a final matching combination according to one embodiment. FIGS. 7 to 9 are flowcharts for explaining the process of displaying matching information according to one embodiment. FIG. 10 is an example diagram of the configuration of a device according to one embodiment. Specific details for implementing the invention

[0019] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0020] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0021] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0022] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or coupled with that other component, or that there may be other components in between.

[0023] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0024] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0025] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0026] The embodiments can be implemented in various forms of products such as personal computers, laptop computers, tablet computers, smartphones, televisions, smart home appliances, intelligent automobiles, kiosks, and wearable devices.

[0027] In the embodiments, the Artificial Intelligence (AI) system is a computer system that implements human-level intelligence, and unlike existing rule-based smart systems, it is a system in which the machine learns and makes judgments on its own. As the AI ​​system improves in recognition rate and becomes capable of understanding sellers' preferences more accurately with use, existing rule-based smart systems are gradually being replaced by deep learning-based AI systems.

[0028] Artificial intelligence technology consists of machine learning and component technologies that utilize machine learning. Machine learning is an algorithmic technology that autonomously classifies and learns the characteristics of input data, while component technologies are technologies that mimic the functions of the human brain, such as cognition and judgment, by utilizing machine learning algorithms like deep learning; these technologies comprise fields such as linguistic understanding, visual understanding, reasoning / prediction, knowledge representation, and motion control.

[0029] The various fields where artificial intelligence technology is applied are as follows. Linguistic understanding refers to the technology of recognizing, applying, and processing human language and text, including natural language processing, machine translation, dialogue systems, question answering, and speech recognition / synthesis. Visual understanding refers to the technology of perceiving and processing objects like human vision, including object recognition, object tracking, image search, person recognition, scene understanding, spatial understanding, and image enhancement. Inference and prediction refers to the technology of logically reasoning and predicting by judging information, including knowledge / probability-based inference, optimization prediction, preference-based planning, and recommendation. Knowledge representation refers to the technology of automatically processing human experiential information into knowledge data, including knowledge construction (data generation / classification) and knowledge management (data utilization). Motion control refers to the technology of controlling autonomous driving of vehicles and the movement of robots, including motion control (navigation, collision, driving) and manipulation control (behavior control).

[0030] Generally, to apply machine learning algorithms to real-world situations, training is performed using a trial-and-error method due to the inherent characteristics of the fundamental methodologies. In particular, deep learning requires hundreds of thousands of iterations. Since it is impossible to execute this in a real physical external environment, training is instead performed through simulations that virtually recreate the actual physical environment on a computer.

[0031] FIG. 1 is a schematic diagram showing the configuration of a system according to one embodiment.

[0032] Referring to FIG. 1, a system according to one embodiment may include a plurality of user terminals (100) and devices (200) capable of communicating with each other through a communication network.

[0033] First, the communication network can be configured regardless of the mode of communication, such as wired or wireless, and can be implemented in various forms to enable communication between servers and between servers and terminals.

[0034] Each of the multiple user terminals (100) can be implemented as a computing device equipped with a communication function, and, for example, can be implemented as a mobile phone, desktop PC, laptop PC, tablet PC, smartphone, etc., but is not limited thereto, and can be implemented as various types of communication devices that can be connected to an external server.

[0035] A plurality of user terminals (100) are terminals used by users who wish to receive matching services between a brand, an artist, and content, and may include a first user terminal (110) used by a first user, a second user terminal (120) used by a second user, etc.

[0036] For convenience of explanation, the operation of the first user terminal (110) is described below, but it is obvious that other user terminals, such as the second user terminal (120), can perform the operation of the first user terminal (110) instead.

[0037] Each of the multiple user terminals (100) can access the device (200) through a web page, an application, etc. provided by the device (200). To this end, an application for providing a matching service between a brand, an artist, and content may be installed on each of the multiple user terminals (100), and the application installed on each of the multiple user terminals (100) may operate in conjunction with the device (200).

[0038] The device (200) may be a private server owned by a person or organization providing a service using the device (200), a cloud server, or a peer-to-peer (P2P) set of distributed nodes. The device (200) may be configured to perform all or part of the computational functions, storage / reference functions, input / output functions, and control functions that a conventional computer possesses. The device (200) may be equipped with at least one artificial intelligence model that performs inference functions.

[0039] The device (200) may include a first artificial intelligence model trained to select recommended artists by analyzing brand information and a second artificial intelligence model trained to select recommended content types, and may be configured to generate matching information between brands, artists and content types based on the inference results of the artificial intelligence models.

[0040] The device (200) can operate in conjunction with a database that stores brand information, artist information, content type information, and past campaign performance data.

[0041] The device (200) may be configured to communicate with a plurality of user terminals (100) via wired or wireless means, control the operation of each of the plurality of user terminals (100), and control which information to display on the screen of each of the plurality of user terminals (100).

[0042] The device (200) is implemented as a server that provides matching services between AI-based brands, artists, and content, and can provide various related services.

[0043] Meanwhile, for convenience of explanation, only the first user terminal (110) and the second user terminal (120) among the plurality of user terminals (100) are shown in FIG. 1, but the number of terminals can vary depending on the embodiment. As long as the processing capacity of the device (200) allows, there is no particular limit to the number of terminals.

[0044] According to one embodiment, the device (200) selects recommended artists and recommended content types based on artificial intelligence, and can provide a matching service between a brand, an artist, and content through the selected recommended artists and recommended content types.

[0045] In the present invention, Artificial Intelligence (AI) refers to a technology that imitates human learning ability, reasoning ability, and perceptual ability, and implements them on a computer, and may include concepts such as machine learning and symbolic logic. Machine Learning (ML) is an algorithmic technology that classifies or learns the characteristics of input data on its own. AI technology can analyze input data as a machine learning algorithm, learn from the results of the analysis, and make judgments or predictions based on the results of the learning. Furthermore, technologies that mimic the functions of the human brain, such as cognition and judgment, by utilizing machine learning algorithms can also be understood as falling within the category of AI. For example, technological fields such as linguistic understanding, visual understanding, reasoning / prediction, knowledge representation, and motion control may be included.

[0046] Machine learning can refer to the process of training neural network models using experience in processing data. It implies that through machine learning, computer software improves its own data processing capabilities. A neural network model is constructed by modeling the correlations between data, and these correlations can be expressed by multiple parameters. A neural network model extracts and analyzes features from given data to derive correlations between them; machine learning can be defined as the process of optimizing the model's parameters by repeating this process. For example, a neural network model can learn the mapping (correlation) between inputs and outputs for data given as input-output pairs. Alternatively, even when only input data is provided, a neural network model can derive regularities between the given data and learn those relationships.

[0047] An artificial intelligence learning model or neural network model can be designed to implement the structure of the human brain on a computer and may include multiple network nodes that have weights and simulate neurons of a human neural network. The multiple network nodes may have interconnected relationships by simulating the synaptic activity of neurons, where neurons exchange signals through synapses. In an artificial intelligence learning model, multiple network nodes may be located in layers of different depths and exchange data according to convolutional connections. The artificial intelligence learning model may be, for example, an Artificial Neural Network (ANN) or a Convolutional Neural Network (CNN). As an embodiment, the artificial intelligence learning model may be machine learned according to methods such as supervised learning, unsupervised learning, and reinforcement learning. Machine learning algorithms for performing machine learning may include Decision Tree, Bayesian Network, Support Vector Machine, Artificial Neural Network, Ada-boost, Perceptron, Genetic Programming, and Clustering.

[0048] Among these, CNNs are a type of multilayer perceptron designed to use minimal preprocessing. CNNs consist of one or more convolutional layers and standard artificial neural network layers stacked on top, additionally utilizing weights and pooling layers. Thanks to this structure, CNNs can fully utilize two-dimensional input data. Compared to other deep learning architectures, CNNs demonstrate good performance in both image and audio fields. CNNs can also be trained using standard backpropagation. CNNs have the advantage of being easier to train than other feedforward artificial neural network techniques and using a small number of parameters.

[0049] Convolutional networks are neural networks comprising sets of nodes with bounded parameters. Many computer vision tasks have been significantly improved, driven by the increased size of available training data and the availability of computational power, combined with algorithmic advancements such as discriminative linear units and dropout training. In the case of massive datasets, such as those available for many tasks today, outfitting is not critical, and increasing the network size improves test accuracy. Optimal utilization of computing resources becomes a limiting factor. To address this, distributed, scalable implementations of deep neural networks can be employed.

[0050] FIG. 2 is a flowchart illustrating the matching process between an AI-based brand, an artist, and content according to one embodiment.

[0051] According to one embodiment, each step illustrated in FIG. 2 may be performed by a device (200), and the device (200) may be configured to provide a matching service between a brand, an artist, and content while communicating with a first user terminal (110). Here, the device (200) may include a first artificial intelligence model trained to select a recommended artist through the analysis of brand information and a second artificial intelligence model trained to select a recommended content type, and may be configured to select a final matching combination based on the inference results of the artificial intelligence models.

[0052] Referring to FIG. 2, first, in step S201, the device (200) may receive a matching request for an artist and content for a first brand from a first user terminal (110). Here, the matching request may refer to a request signal transmitted to the device (200) through the first user terminal (110) to receive recommendations for an artist and content type suitable for the first brand.

[0053] For example, when an advertiser, brand operator, marketing manager, or agency manager selects a campaign creation menu, an artist recommendation menu, or a content planning menu for a specific brand on the screen of the first user terminal (110), the first user terminal (110) can transmit a matching request for the brand to the device (200). At this time, the matching request may include at least one of the following information: brand name, campaign purpose, budget range, target age group, target gender, preferred image, desired content channel, timing of execution, etc.

[0054] That is, the device (200) can initiate a matching procedure to recommend an ad execution combination suitable for the first brand by receiving a matching request for an artist and content for the first brand from the first user terminal (110).

[0055] In step S202, the device (200) can obtain first brand information, which is information related to the first brand. Here, the first brand information may be defined as information representing the characteristics, marketing objectives, and past performance of the first brand, and may be obtained through information stored in the database of the device (200), information input from the first user terminal (110), information received via linkage from an external platform, or a combination thereof.

[0056] According to one embodiment, the first brand information may include the industry group to which the brand belongs, brand image, target consumer group, price range, product or service category, history of past advertising campaigns, views of past campaigns, click-through rate, conversion rate, revenue contribution, preferred artist type, preferred content format, country or region information, etc. Additionally, the first brand information may be configured in the form of text information, categorical information, numerical information, image information, or composite data combining these.

[0057] That is, the device (200) can collect or look up various attribute information about the first brand and secure the first brand information to be used as input to an artificial intelligence model thereafter.

[0058] In step S203, the device (200) can generate a first input signal by encoding the first brand information. Here, encoding may refer to the process of converting the first brand information into data in a format that can be processed by an artificial intelligence model.

[0059] For example, the device (200) can generate a first input signal by converting text data included in the first brand information into an embedding vector, converting categorical data into one-hot encoding or index mapping methods, and normalizing or standardizing numerical data and combining them. In this case, if image information is included, the device (200) can generate an image feature vector through a feature extraction model and include it in the first input signal.

[0060] According to one embodiment, the first input signal may be a single vector representing the current characteristics of a brand, or it may be in the form of a combination of multiple attribute vectors.

[0061] That is, the device (200) can convert the first brand information into a form that can be interpreted by a machine learning model or a neural network model, and generate a first input signal that reflects brand characteristics.

[0062] In step S204, the device (200) may input a first input signal to a first artificial intelligence model trained to select a recommended artist through the analysis of brand information. Here, the first artificial intelligence model may be a model trained to infer an artist suitable for a specific brand by learning the correspondence relationship between brand information and the artist.

[0063] For example, the first artificial intelligence model may be pre-trained using past brand-artist collaboration history, advertising performance data, fandom characteristics by artist, content response data, and brand image fit data as training data. In this case, the first artificial intelligence model may be implemented as an artificial neural network, a deep neural network, a transformer-based model, a classification model, a ranking model, or a combination thereof.

[0064] The device (200) provides a first input signal to a first artificial intelligence model and can obtain a suitability, ranking, or selection probability for each of a plurality of candidate artists as an inference result. That is, the device (200) can input a first input signal representing brand characteristics into the first artificial intelligence model to infer a recommended artist suitable for the first brand.

[0065] In step S205, when the device (200) selects the recommended artist of the first brand as the first artist through the first input signal, it may obtain a first output signal representing the first artist from the first artificial intelligence model. Here, the first output signal may be defined as a signal including identification information, recommendation score, ranking information, or selection result of the first artist, and for example, the first output signal may include at least one of the first artist's unique identification number, name, affiliated category, representative image, suitability score, and recommendation reason information.

[0066] According to one embodiment, the first artificial intelligence model may select an artist with the highest recommendation score among a plurality of candidate artists as the recommended artist, or select an artist with the highest ranking among artists satisfying a preset threshold score or higher as the recommended artist. In this case, the output format of the first artificial intelligence model may be a probability value, a class label, an embedding similarity score, a ranking result, or a combination thereof. A detailed explanation of the process of selecting a recommended artist in the first artificial intelligence model will be described later with reference to FIG. 3.

[0067] That is, the device (200) can identify a recommended artist suitable for the first brand by obtaining a first output signal containing the inference result of the first artificial intelligence model.

[0068] According to one embodiment, the first artificial intelligence model may be a model trained to calculate a matching score between a brand and an artist by learning the correspondence relationship between brand information and artist information. Here, the first artificial intelligence model may be configured to analyze input data representing brand characteristics and input data representing artist characteristics together to infer the degree of fit between the two data.

[0069] Specifically, the first artificial intelligence model may be implemented as a neural network structure that converts brand information and artist information into feature vectors, respectively, and then combines the feature vectors to calculate a matching score. For example, the first artificial intelligence model may include a first encoder that receives brand information and a second encoder that receives artist information, and the first encoder and the second encoder may be configured to generate a brand feature vector and an artist feature vector, respectively. Subsequently, the first artificial intelligence model may include a matching network that calculates a matching score by combining or comparing the brand feature vector and the artist feature vector.

[0070] Additionally, the first artificial intelligence model may be implemented as a multilayer perceptron (MLP), a deep neural network (DNN), a transformer-based neural network, an embedding-based recommendation model, or a combination thereof. For example, the first artificial intelligence model may be configured to receive feature vectors representing brand image, target audience, and campaign performance, and feature vectors representing artist image, fan base, and advertising performance as input, learn the correlation between the two vectors, and output a matching score representing the suitability between the brand and the artist based on this.

[0071] In the learning process of the first artificial intelligence model, the device (200) may use past advertising campaign data, collaboration history data between a brand and an artist, advertising performance data, and user response data as learning data. For example, the device (200) may use the data as positive learning data if the collaboration between a specific brand and a specific artist achieved a high participation rate or conversion rate in an actual advertising campaign, and may use the data as negative learning data if the advertising performance was low or the collaboration did not take place. At this time, the first artificial intelligence model may be learned using at least one of supervised learning, ranking learning, or reinforcement learning methods, and the model parameters may be repeatedly updated during the learning process to minimize the difference between the matching score and the actual advertising performance.

[0072] In other words, the first artificial intelligence model can be trained to predict artists with high potential for advertising performance for a specific brand by learning the relationship between brand characteristic data and artist characteristic data.

[0073] In addition to the learning method described above, the first artificial intelligence model may also be trained through various learning methods of the known artificial intelligence model.

[0074] In step S206, the device (200) may determine the recommended artist of the first brand as the first artist based on the first output signal. Here, determining the recommended artist may mean determining the final artist to be used as a matching target with the first brand from among a plurality of candidate artists.

[0075] For example, the device (200) can determine the first artist by analyzing recommendation scores, rankings, and reliability information included in the first output signal, and can make a final determination by additionally reflecting preset filter conditions, such as activity area, contract availability, schedule availability, and advertising restriction industry, as needed.

[0076] The device (200) may not only determine a single artist, but also determine the top artist among multiple recommended artists as the first artist while maintaining the remaining artists as a candidate group.

[0077] That is, the device (200) can interpret the first output signal and determine the recommended artist of the first brand to be used for generating actual matching information as the first artist.

[0078] In step S207, the device (200) may input a first input signal to a second artificial intelligence model trained to select a recommended content type by analyzing brand information. Here, the second artificial intelligence model may be a model trained to infer a content type suitable for a specific brand by learning the correspondence relationship between brand information and content type.

[0079] For example, the second artificial intelligence model can be pre-trained using training data such as content formats used in past brand campaigns, campaign performance data, content response data such as engagement rates by content type, views and conversion rates, and suitability data between the brand image and the content presentation method. In this case, the second artificial intelligence model can be implemented as an artificial neural network, a deep neural network, a transformer-based model, a classification model, a ranking model, or a combination thereof.

[0080] The device (200) provides a first input signal to a second artificial intelligence model and can obtain a suitability, ranking, or selection probability for each of a plurality of candidate content types as an inference result. Here, the content type may refer to an expression format or production format of content for conveying a brand message, and may include, for example, short-form video content, vlog-type content, product experience-type content, interview-type content, challenge-type content, or live broadcast content.

[0081] That is, the device (200) can input a first input signal representing brand characteristics into a second artificial intelligence model to infer a type of recommended content suitable for the first brand.

[0082] In step S208, when the device (200) selects the recommended content type of the first brand as the first type through the first input signal, it may obtain a second output signal representing the first type from the second artificial intelligence model. Here, the second output signal may be defined as a signal including identification information of the first type, content format information, recommendation score, ranking information, or selection result, and for example, the second output signal may include a format code classified as short-form video, vlog, experiential review, interview-type advertisement, etc., along with a suitability score or priority for each format.

[0083] According to one embodiment, the second artificial intelligence model may obtain a score for each of a plurality of candidate content types and select the content type with the highest score as the recommended content type, and may also select the recommended content type by additionally considering platform suitability, brand budget, production period, campaign objectives, etc., as necessary. At this time, the output format of the second artificial intelligence model may be a probability value, a class label, an embedding similarity score, a ranking result, or a combination thereof. A detailed explanation of the process of selecting the recommended content type in the second artificial intelligence model will be described later with reference to FIG. 4.

[0084] That is, the device (200) can obtain a second output signal including the output result of the second artificial intelligence model and determine the type of recommended content suitable for the first brand.

[0085] According to one embodiment, the second artificial intelligence model may be a model trained to calculate a matching score between a brand and a content type by learning the correspondence relationship between brand information and content type information. Here, the second artificial intelligence model may be configured to analyze input data representing brand characteristics and input data representing content type characteristics together to infer the degree of fit between the two data.

[0086] Specifically, the second artificial intelligence model may be implemented as a neural network structure that converts brand information and content type information into feature vectors, respectively, and then combines the feature vectors to calculate a matching score. For example, the second artificial intelligence model may include a first encoder that receives brand information and a second encoder that receives content type information, and the first encoder and the second encoder may be configured to generate a brand feature vector and a content type feature vector, respectively. Subsequently, the second artificial intelligence model may include a matching network that calculates a matching score by combining or comparing the brand feature vector and the content type feature vector.

[0087] Additionally, the second AI model can be implemented as a multilayer perceptron (MLP), a deep neural network (DNN), a transformer-based neural network, an embedding-based recommendation model, or a combination thereof. For example, the second AI model may be configured to receive feature vectors representing brand image, target audience, and campaign performance, and feature vectors representing content image, content main audience, and content advertising performance as input, learn the correlation between the two vectors, and output a matching score indicating the suitability between the brand and the content type based on this.

[0088] In the learning process of the second artificial intelligence model, the device (200) may use past advertising campaign data, application history data between brands and content types, content advertising performance data, and user response data as learning data. For example, the device (200) may use the data as positive learning data if a specific brand achieves high views, engagement rates, or conversion rates in an advertising campaign using a specific content type, and may use the data as negative learning data if advertising performance is low. At this time, the second artificial intelligence model may be learned using at least one of supervised learning, ranking learning, or reinforcement learning methods, and model parameters may be repeatedly updated to minimize the difference between the matching score calculated during the learning process and the actual advertising performance.

[0089] In other words, the second artificial intelligence model can be trained to predict content types with high potential for advertising delivery effectiveness and advertising performance for a specific brand by learning the relationship between brand characteristic data and content type characteristic data.

[0090] In addition to the learning method described above, the second artificial intelligence model may also be trained through various learning methods of the known artificial intelligence model.

[0091] In step S209, the device (200) can determine the recommended content type of the first brand as the first type based on the second output signal. Here, the determination of the recommended content type may mean determining the final content format to be used in the actual matching combination among a plurality of candidate content formats.

[0092] For example, the device (200) can determine a first type by comparing the score, rank, or probability included in the second output signal, and additionally, it may be configured to prioritize a specific type among the same output results depending on whether the brand's campaign objective is to increase awareness or to convert sales.

[0093] The device (200) may determine the first type as a single type, or may determine multiple types including a main type and an auxiliary type.

[0094] That is, the device (200) can interpret the second output signal and determine the type of recommended content of the first brand to be used for generating actual matching information as the first type.

[0095] In step S210, the device (200) may generate first matching information by matching a first brand, a first artist, and a first type of content. Here, the first matching information may be defined as information representing the result of a combination of an artist and a content type suitable for a specific brand, and for example, the first matching information may include identification information of the first brand, identification information of the first artist, identification information of the first type, recommendation scores for each combination, expected performance indicators, target suitability, expected reach, expected engagement rate, expected conversion rate, recommendation reason information, production guide information, or campaign execution plan information.

[0096] The device (200) may not merely combine the first artist and the first type, but may also generate a prediction result regarding the extent to which the combination has performed in past similar brand campaigns and include this result in the first matching information. At this time, the device (200) may be configured to select a final matching combination by comprehensively considering the suitability between the first brand and the first artist, the suitability between the first brand and the first type, and the suitability between the first artist and the first type. A detailed explanation of the process of selecting the final matching combination will be provided later with reference to FIGS. 5 and 6.

[0097] That is, the device (200) can generate first matching information that integrally reflects a combination of brand, artist, and content type, thereby providing a result that can be used for actual advertising or branded content planning.

[0098] According to one embodiment, the device (200) can select a combination predicted to have the highest advertising performance among a plurality of candidate combinations as the final matching combination by considering together the suitability between the first brand and the first artist, the suitability between the first brand and the first type, and the suitability of content expression between the first artist and the first type.

[0099] In step S211, the device (200) may transmit first matching information to the first user terminal (110). Here, the transmission of the first matching information may mean the process of providing the recommendation result generated by the device (200) in a form that the user can verify.

[0100] For example, the device (200) may provide a user interface on the screen of the first user terminal (110) that includes a recommended artist for the first brand, a recommended content type, expected advertising performance, grounds for recommendation, similar campaign examples, and a follow-up button. In this case, the first matching information may be provided in the form of an electronic document, a file in the form of a proposal, a dashboard screen, or an API response, in addition to simple display.

[0101] The device (200) transmits the first matching information to the first user terminal (110), thereby enabling the user to quickly identify a suitable ad combination based on the analysis results of the artificial intelligence model and subsequently proceed with the contract, execute a campaign, or compare additional candidates. A detailed description of the process of transmitting the first matching information to the first user terminal (110) and displaying the first matching information will be described later with reference to FIGS. 7 to 9.

[0102] That is, the device (200) can provide the user with artist and content matching results for the first brand, thereby enabling data-based advertising decision-making.

[0103] After generating the first matching information, the device (200) may collect the user's selection results, actual campaign execution results, and performance feedback and use them for retraining the first artificial intelligence model and the second artificial intelligence model.

[0104] For example, if the first artist and first type of content recommended for a specific brand are actually adopted and the participation rate or conversion rate of the campaign is high, the device (200) can reflect the result as positive learning data to improve the accuracy of recommendations for similar brands thereafter. Conversely, if the recommendation result is not adopted or the performance is poor, the device (200) can reflect this as correction data to adjust the matching logic. Through this, the device (200) can continuously improve the matching performance between brands, artists, and content based on data accumulated during the process of repeated advertising execution.

[0105] Accordingly, by utilizing an artificial intelligence model to automatically select artists and content types suitable for a brand based on the analysis of brand information and generating matching information by combining them, data-driven optimal matching can be provided during the advertising or branded content planning process, thereby improving marketing efficiency and the potential for campaign performance.

[0106] FIG. 3 is a flowchart for explaining the process of selecting a recommended artist according to one embodiment.

[0107] According to one embodiment, each step illustrated in FIG. 3 can be performed in conjunction with step S205, and the first artificial intelligence model can calculate a matching score between each of the first brand and a plurality of artists, and select the recommended artist most suitable for the first brand by comparing the calculated matching scores. Here, the first artificial intelligence model may be a model trained to calculate a matching score representing the degree of suitability between the brand and the artist by learning the correspondence relationship between brand information and artist information.

[0108] Referring to FIG. 3, first, in step S301, the first artificial intelligence model can identify first brand information based on the first input signal. Here, the first brand information may be brand-related information that can be restored or interpreted from the first input signal.

[0109] In step S302, the first artificial intelligence model can identify the first brand's brand image, target customer base, and brand campaign performance based on the first brand information. Here, brand image may refer to the identity, emotional atmosphere, or market positioning that the brand intends to convey to consumers, and may be categorized, for example, into a luxurious image, an eco-friendly image, a young and trendy image, a popular image, or a professional image. Additionally, the target customer base may refer to the first brand's main consumer group and may be categorized by age group, gender, interests, purchasing power, region, or lifestyle. Furthermore, brand campaign performance refers to information representing the results of past advertising or branded content campaigns conducted by the brand, and may include views, click-through rates, engagement rates, conversion rates, sales contribution, or brand awareness growth rates.

[0110] In other words, the first AI model extracts key attribute information included in the first brand information, and can identify brand image, target audience, and brand campaign performance, respectively, as brand reference information to be used for similarity comparison with the artist.

[0111] In step S303, the first artificial intelligence model may select a first artist, which is one of a plurality of pre-registered artists, as the subject of analysis. Here, the plurality of artists may include advertising models, influencers, celebrities, creators, or content producers that are pre-registered in the database of the device (200), and, for example, each of the plurality of artists may be stored in association with their name, field of activity, image characteristics, fan base characteristics, past advertising participation history, and advertising performance.

[0112] The first AI model may select the artist corresponding to the current analysis turn as the first artist in order to calculate matching scores for each of the multiple artists sequentially or in parallel. That is, the first AI model may designate one of the multiple candidate artists as the subject of analysis and prepare to calculate the suitability with the first brand.

[0113] In step S304, the first artificial intelligence model can obtain first artist information, which is information related to the first artist. Here, the first artist information may be obtained through information stored in the database of the device (200), information received via linkage from an external platform, or a combination thereof. For example, the first artist information may include the first artist's name, activity genre, representative image, fan base characteristics, past advertising participation history, advertising performance, main activity channels, content reaction data, country or region information, etc. Additionally, the first artist information may be configured in the form of text information, categorical information, numerical information, image information, or composite data combining these.

[0114] In other words, the first artificial intelligence model can obtain first artist information as a criterion for analyzing the suitability with the first brand.

[0115] In step S305, the first artificial intelligence model can identify the first artist's artist image, fan base, and artist advertising performance based on the first artist's information. Here, the artist image may refer to the visual or emotional impression the artist forms with the public, and may be categorized, for example, into a sophisticated image, a friendly image, a luxurious image, a dynamic image, or a trendy image. Additionally, the fan base may refer to a group of users who prefer or continuously consume the first artist, and may be categorized by age group, gender, interests, activity platform, region, or consumption tendency. Furthermore, artist advertising performance refers to information indicating the performance of past advertisements or sponsored content in which the artist participated, and may include views, engagement rate, conversion rate, sales contribution, comment response rate, or number of shares.

[0116] In other words, the first artificial intelligence model extracts attribute information included in the first artist information and can verify the artist image, fan base, and artist advertising performance, respectively, as artist standard information for comparison with the first brand.

[0117] In step S306, the first artificial intelligence model can calculate the similarity between the brand image of the first brand and the artist image of the first artist as the first similarity. Here, the first similarity can be defined as a value representing the degree of congruence between the identity or atmosphere pursued by the brand and the image characteristics possessed by the artist.

[0118] For example, the first artificial intelligence model may vectorize the brand image and the artist image, respectively, and then calculate a first similarity using at least one of cosine similarity, a Euclidean distance-based score, an inner product-based score, or a pre-trained embedding similarity calculation method. In this case, a higher value of the first similarity may be interpreted as indicating that the images of the brand and the artist are more similar or harmonious.

[0119] In other words, the first artificial intelligence model can quantify the image aspect fit between the first brand and the first artist and calculate the first similarity.

[0120] In step S307, the first artificial intelligence model can calculate the similarity between the target customer base of the first brand and the fan base of the first artist as the second similarity. Here, the second similarity may be a value representing the degree of congruence between the consumer group that the brand intends to reach and the fan group over which the first artist's actual influence extends.

[0121] For example, the first artificial intelligence model can calculate the second similarity by comparing the age distribution, gender distribution, interest distribution, regional distribution, or consumption propensity distribution of the target customer base and the fan base, respectively.

[0122] Additionally, the first artificial intelligence model can calculate a second similarity through at least one of a category-specific match rate, probability distribution similarity, inter-cluster distance, or embedding-based similarity calculation method. In this case, a higher value of the second similarity can be interpreted as indicating that the main customer base of the first brand and the fan base of the first artist match better.

[0123] In other words, the first artificial intelligence model quantifies the degree of overlap of the reachable targets and can represent the target fit between the first brand and the first artist as the second similarity.

[0124] In step S308, the first artificial intelligence model can calculate the third similarity as the similarity between the brand campaign performance of the first brand and the artist advertising performance of the first artist. Here, the third similarity may be a value representing the similarity between the campaign performance characteristics achieved by the first brand in the past and the performance characteristics shown by the first artist in past advertisements.

[0125] For example, the first AI model can calculate the third similarity by comparing the indicators that showed high performance in the first brand's past campaigns with the indicators that showed high performance during the first artist's past advertising participation.

[0126] More specifically, the first AI model can generate vectors based on indicators such as views, engagement rate, click-through rate, conversion rate, sales contribution, or brand favorability growth rate, and calculate the third similarity by determining the similarity between the vectors. In this case, a higher value for the third similarity can be interpreted as indicating that the advertising performance pattern expected by the first brand is more similar to the advertising performance pattern actually demonstrated by the first artist.

[0127] In other words, the first artificial intelligence model can calculate the performance fit between the first brand and the first artist as the third similarity based on the correspondence relationship of past performance data.

[0128] In step S309, the first artificial intelligence model can calculate a matching score between the first brand and the first artist by summing the first similarity, the second similarity, and the third similarity. Here, the matching score may be a numerical value representing the overall suitability between the first brand and the first artist.

[0129] For example, the first AI model may simply sum the first, second, and third similarities with equal weights, or it may calculate the matching score using a weighted summation method in which different weights are assigned based on the importance of brand image, the importance of target match, and the importance of past performance fit. In this case, the first AI model may normalize or standardize each similarity to the same scale before summing.

[0130] In other words, the first artificial intelligence model can calculate a matching score between the first brand and the first artist by comprehensively reflecting image suitability, fan base suitability, and performance suitability.

[0131] The first artificial intelligence model can calculate matching scores between the first brand and multiple artists, respectively, using the same method as calculating the matching score between the first brand and the first artist. At this time, the calculation of matching scores for each of the first brand and multiple artists can be achieved by repeatedly performing steps S303 through S309.

[0132] In step S310, when the first artificial intelligence model calculates matching scores for the first brand and multiple artists respectively, if it is confirmed that the matching score between the first brand and the first artist is the highest as a result of comparing the matching scores between the first brand and multiple artists, the recommended artist of the first brand can be selected as the first artist.

[0133] For example, the first artificial intelligence model can calculate a matching score for each of multiple artists, sort the matching scores in descending order, and select the artist with the highest score as the recommended artist for the first brand.

[0134] In other words, the first artificial intelligence model compares the matching scores for each of the multiple candidate artists and can select the first artist, who is the most suitable artist for the first brand, as the recommended artist.

[0135] Accordingly, by calculating a matching score based on the similarity between brand image, target audience, and campaign performance and artist image, fan base, and advertising performance, it becomes possible to provide data-driven artist recommendations that simultaneously reflect brand characteristics and advertising performance potential.

[0136] FIG. 4 is a flowchart for explaining the process of selecting a recommended content type according to one embodiment.

[0137] According to one embodiment, each step illustrated in FIG. 4 can be performed in conjunction with step S208, and the second artificial intelligence model can calculate a matching score between each of the first brand and a plurality of content types, and select the most suitable recommended content type for the first brand by comparing the calculated matching scores. Here, the second artificial intelligence model may be a model trained to calculate a matching score indicating the degree of suitability between the brand and the content type by learning the correspondence relationship between brand information and content type information.

[0138] Referring to FIG. 4, first, in step S401, the second artificial intelligence model can verify the first brand information based on the first input signal. Here, the first brand information may be brand-related information that can be restored or interpreted from the first input signal.

[0139] In the S402 stage, the second AI model can verify the first brand's brand image, target customer base, and brand campaign performance based on the first brand's information.

[0140] In other words, the second AI model extracts key attribute information included in the first brand information, and can identify brand image, target audience, and brand campaign performance, respectively, as brand reference information to be used for comparing similarity with content types.

[0141] In step S403, the second artificial intelligence model may select a first type, which is one of a plurality of pre-registered content types, as the target for analysis. Here, the plurality of content types may include content formats or production formats pre-registered in the database of the device (200), and may include, for example, short-form video content, vlog-type content, product experience-type content, interview-type content, challenge-type content, live broadcast content, or story-type content.

[0142] In order to calculate matching scores for each of multiple content types sequentially or in parallel, the second AI model may select the content type corresponding to the current analysis turn as the first type. That is, the second AI model may designate one of the multiple candidate content types as the analysis target and prepare to calculate the suitability with the first brand.

[0143] In step S404, the second artificial intelligence model can obtain first content type information, which is information related to the first type. Here, the first content type information can be obtained through information stored in the database of the device (200), information received via linkage from an external content platform, or a combination thereof. For example, the first content type information may include content format names, content image characteristics, main audiences, past advertising application cases, content advertising performance, major publishing platforms, content response data, or country and region information. Additionally, the first content type information may be configured in the form of text information, categorical information, numerical information, image information, or composite data combining these.

[0144] In other words, the second artificial intelligence model can obtain information on the first content type as a criterion for analyzing the suitability with the first brand.

[0145] In step S405, the second AI model can identify the content image, the content's primary audience, and the content advertising performance of the first type based on the information of the first content type. Here, the content image may refer to the mode of expression or the atmosphere of the content conveyed by the content type, and may be classified, for example, as an informational image, an entertainment image, an experience-oriented image, an emotional story image, or a challenge image. Additionally, the content's primary audience may refer to the user group that primarily consumes the content type, and may be classified by age group, gender, interests, activity platform, region, or consumption patterns. Furthermore, content advertising performance may refer to the performance achieved when a specific content type is utilized as advertising or sponsored content, and may include views, engagement rate, click-through rate, conversion rate, number of shares, or brand response.

[0146] In other words, the second AI model extracts attribute information included in the first content type information and can identify the content image, content target audience, and content advertising performance, respectively, as content reference information for comparison with the first brand.

[0147] In step S406, the second artificial intelligence model can calculate the similarity between the brand image of the first brand and the content image of the first type as the fourth similarity. Here, the fourth similarity can be defined as a value representing the degree of congruence between the identity or atmosphere pursued by the brand and the mode of expression conveyed by the content type.

[0148] For example, the second AI model may vectorize the brand image and the content image, respectively, and then calculate a fourth similarity using at least one of cosine similarity, a Euclidean distance-based score, an inner product-based score, or a pre-trained embedding similarity calculation method. In this case, a higher value of the fourth similarity may be interpreted as indicating that the images of the brand and the content type are more similar or harmonious.

[0149] That is, the second artificial intelligence model can quantify the image aspect fit between the first brand and the first type and calculate the fourth similarity.

[0150] In step S407, the second AI model can calculate the similarity between the target customer base of the first brand and the main audience of the first type of content as the fifth similarity. Here, the fifth similarity may be a value representing the degree of congruence between the consumer group that the brand intends to reach and the audience group that actually consumes a specific type of content.

[0151] For example, the second artificial intelligence model can calculate the fifth similarity by comparing the age distribution, gender distribution, interest distribution, regional distribution, or consumption propensity distribution of the target customer base and the main content audience, respectively. Additionally, the fifth similarity can be calculated through at least one of the following methods: category-specific match rate, probability distribution similarity, inter-cluster distance, or embedding-based similarity calculation method.

[0152] In other words, the second AI model quantifies the degree of overlap between viewer groups and can represent the target fit between the first brand and the first type as the fifth similarity.

[0153] In step S408, the second AI model can calculate the similarity between the brand campaign performance of the first brand and the performance of the first type of content advertisement as the sixth similarity. Here, the sixth similarity may be a value representing the similarity between the performance characteristics achieved by the first brand in past campaigns and the performance characteristics shown by a specific content type in past advertisements.

[0154] For example, the second artificial intelligence model can generate vectors based on indicators such as views, engagement rate, click-through rate, conversion rate, sales contribution, or brand response growth rate, and calculate the similarity between the vectors to obtain the sixth similarity.

[0155] In other words, the second artificial intelligence model can calculate the performance fit between the first brand and the first type as the sixth similarity based on the correspondence of past performance data.

[0156] In step S409, the second artificial intelligence model can calculate a matching score between the first brand and the first type by summing the fourth similarity, the fifth similarity, and the sixth similarity. Here, the matching score may be a numerical value representing the overall suitability between the first brand and the first type.

[0157] For example, the second AI model can simply sum the fourth, fifth, and sixth similarities with equal weights, or it can calculate the matching score using a weighted summation method that assigns different weights based on the importance of brand image fit, audience match, and advertising performance fit. Additionally, each similarity can be normalized or standardized to the same scale before summing.

[0158] In other words, the second AI model can calculate a matching score between the first brand and the first type by comprehensively reflecting content image suitability, audience suitability, and advertising performance suitability.

[0159] The second artificial intelligence model can calculate matching scores between the first brand and multiple content types, respectively, using the same method as calculating the matching score between the first brand and the first type. In this case, the calculation of matching scores for each of the first brand and multiple content types can be achieved by repeatedly performing steps S403 through S409.

[0160] In step S410, when the second artificial intelligence model calculates matching scores for the first brand and multiple content types respectively, if it is confirmed that the matching score between the first brand and the first type is the highest as a result of comparing the matching scores between the first brand and the multiple content types, the recommended content type of the first brand can be selected as the first type.

[0161] For example, the second artificial intelligence model can calculate a matching score for each of multiple content types, sort the matching scores in descending order, and select the content type with the highest score as the recommended content type for the first brand.

[0162] In other words, the second AI model compares the matching scores for each of the multiple candidate content types and can select Type 1, which is the content type most suitable for the first brand, as the recommended content type.

[0163] Accordingly, by calculating a matching score based on the similarity between brand image, target audience, and campaign performance and content image, content main audience, and content advertising performance, it becomes possible to provide data-driven content type recommendations that simultaneously reflect brand characteristics and content delivery effects.

[0164] FIGS. 5 and 6 are flowcharts for explaining the process of selecting a final matching combination according to one embodiment.

[0165] According to one embodiment, each step illustrated in FIGS. 5 to 6 can be performed in conjunction with step S210, and the device (200) can set a combination of a first artist and a first type pre-selected for a first brand as a reference combination, expand a group of candidates including artists and content types similar to the reference combination, generate multiple combinations, and predict advertising performance for each of the multiple combinations to select a final matching combination. Here, the device (200) can be configured to form a group of candidates using the matching score between the brand and the artist calculated in FIG. 3 and the matching score between the brand and the content type calculated in FIG. 4, and to finally generate first matching information by comparing the performance prediction results for multiple combinations.

[0166] Referring to FIGS. 5 and 6, first, in step S501, the device (200) can predict advertising performance with the first brand for a combination of the first artist and the first type and calculate a first performance prediction score. Here, the combination of the first artist and the first type may be a combination of a recommended artist and a recommended content type determined in steps S206 and S209, respectively. Additionally, advertising performance may refer to performance expected when the combination is applied to an actual advertising or branded content campaign, and may include, for example, a prediction result calculated based on views, engagement rate, click-through rate, conversion rate, sales contribution, brand responsiveness, or a combination thereof.

[0167] The device (200) can calculate a first performance prediction score through a performance prediction model that uses brand information of the first brand, first artist information, and first content type information as input values. That is, the device (200) can quantify the potential for advertising performance for a standard combination consisting of the first artist and the first type and calculate it as a first performance prediction score. Here, the first performance prediction score can be used as a standard performance value for the initial recommended combination and can be utilized as a standard value for comparison with the performance prediction results for subsequent candidate combinations.

[0168] Specifically, the device (200) can calculate a performance prediction score using a performance prediction artificial intelligence model that uses first brand information, first artist information, and first content type information as input values. Here, the performance prediction artificial intelligence model is a model trained on past advertising campaign data and can be configured to predict advertising performance that may occur when a specific combination of brand, artist, and content type is applied to an actual advertising campaign.

[0169] For example, the device (200) can convert the brand image of the first brand, target customer base, campaign performance history, artist image of the first artist, fan base characteristics, and advertising performance history, as well as the image of the first type of content, content main audience, and content advertising performance history into feature vectors and input them into a performance prediction artificial intelligence model. At this time, the feature vectors may be composed in a combined form of categorical data, numerical data, and embedding data.

[0170] The device (200) can combine (concatenation) the brand feature vector, the artist feature vector, and the content type feature vector into a single input vector by weighting and combining them, and then input it into a performance prediction artificial intelligence model.

[0171] According to one embodiment, the performance prediction artificial intelligence model may include a multilayer neural network structure and may be composed of an input layer, a plurality of hidden layers, and an output layer. In the input layer, an input vector combining a brand feature vector, an artist feature vector, and a content type feature vector may be input, and in the hidden layer, a non-linear transformation may be performed on the input vector to learn the interrelationships between the brand, the artist, and the content. In the output layer, a predicted value representing the advertising performance for a specific combination may be output.

[0172] For example, the advertising performance prediction value may include at least one of the expected number of views, expected engagement rate, expected click-through rate, expected conversion rate, or expected revenue contribution. The device (200) may calculate a performance prediction score by applying a preset weight to each prediction value and performing a weighted summation. At this time, the performance prediction score may be calculated as follows.

[0173] Performance Prediction Score = (Predicted Views × w1) + (Predicted Engagement × w2) + (Predicted Click-through Rate × w3) + (Predicted Conversion Rate × w4)

[0174] Here, w1 through w4 may be weights set according to the purpose of the advertising campaign. For example, if the goal is to increase brand awareness, higher weights may be assigned to views and engagement rates, and if the goal is to increase sales conversions, higher weights may be assigned to click-through rates and conversion rates.

[0175] Additionally, the performance prediction AI model can be trained using combinations of brands, artists, and content types used in past advertising campaigns and actual advertising performance data as training data. For example, the device (200) can train the performance prediction AI model using a supervised learning method by using actual data such as views, engagement rate, click-through rate, conversion rate, and sales contribution for combinations used in past advertising campaigns as label data.

[0176] In this way, the device (200) can calculate a performance prediction score that considers the actual potential for advertising performance beyond simple matching score-based recommendations by using an artificial intelligence model that predicts advertising performance using a combination of brand, artist, and content type as input values.

[0177] In step S502, the device (200) can check the matching score with the first brand for each of the multiple artists. Here, the matching score for each of the multiple artists may be the score calculated in step S309 or a score calculated and stored in the same manner. For example, the device (200) can query or recalculate the matching score from the database that comprehensively reflects the image suitability, target suitability, and performance suitability between the first brand and each of the multiple artists. That is, the device (200) can check the score value indicating the degree of suitability of each of the multiple artists based on the first brand.

[0178] In step S503, the device (200) may set a matching score between the first brand and the first artist as a first reference score. Here, the first reference score may be defined as a reference value representing the suitability of the first artist already recommended for the first brand. For example, the device (200) may use the matching score of the first artist selected as the recommended artist in FIG. 3 as a reference value to expand the candidate group to include artists of a similar level. That is, the device (200) may set a score representing the brand suitability of the first artist as the first reference score and use it as a reference value for selecting candidate artists.

[0179] In step S504, the device (200) may classify artists among a plurality of artists whose difference between each artist's matching score and a first reference score is less than or equal to a first reference value as a first candidate group. Here, the first reference value refers to an allowable range that can be recognized as being similar to the matching score of the first artist, and may be set differently depending on the embodiment. At this time, the first candidate group may be a set of a plurality of artists whose matching score with the first brand is confirmed to be similar to the first reference score while excluding the first artist.

[0180] For example, the device (200) can calculate an absolute value by subtracting a first reference score from the matching score of each artist, and if the calculated absolute value is less than or equal to the first reference value, the artist can be classified as a first candidate group. At this time, the first reference value may be a fixed value or may be dynamically set as a certain ratio of the first reference score. That is, the device (200) can select multiple artists with a similar level of suitability to the first artist as a first candidate group.

[0181] In step S505, the device (200) can check the matching score with the first brand for each of the multiple content types. Here, the matching score for each of the multiple content types may be the score calculated in step S409 or a score calculated and stored in the same manner. For example, the device (200) can query or recalculate the matching score from the database that comprehensively reflects the content image suitability, audience suitability, and advertising performance suitability between the first brand and each of the multiple content types. That is, the device (200) can check the score value indicating the degree of suitability of each of the multiple content types based on the first brand.

[0182] In step S506, the device (200) may set a matching score between the first brand and the first type as a second reference score. Here, the second reference score may be defined as a reference value indicating the suitability of the first type already recommended for the first brand. For example, the device (200) may use the matching score of the first type selected as the recommended content type in FIG. 4 as a reference value to expand the candidate group of content types of a similar level. That is, the device (200) may set a score indicating the brand suitability of the first type as the second reference score and use it as a reference value for selecting candidate content types.

[0183] In step S507, the device (200) may classify content types among a plurality of content types into a second candidate group in which the difference between the matching score of each content type and the second reference score is less than or equal to the second reference value. Here, the second reference value refers to an allowable range that can be recognized as being similar to the matching score of the first type, and may be set differently depending on the embodiment. At this time, the second candidate group may be a set of a plurality of content types in which the matching score with the first brand is confirmed to be similar to the second reference score while excluding the first type.

[0184] For example, the device (200) calculates an absolute value by subtracting a second reference score from the matching score of each content type, and if the absolute value is less than or equal to the second reference value, the content type can be classified as a second candidate group. At this time, the second reference value may be a fixed value or may be dynamically set as a certain ratio of the second reference score. That is, the device (200) can select multiple content types that have a similar level of suitability with the first type as a second candidate group.

[0185] According to one embodiment, the first threshold and the second threshold may each be set as values ​​representing an allowable error range for selecting candidates having a level of suitability similar to the reference artist and reference content type. In this case, the first threshold and the second threshold may be set as fixed threshold values ​​or as relative ratio values ​​to the reference score.

[0186] For example, the device (200) can set a value corresponding to a certain percentage of the first reference score as the first reference value, and classify artists included within a certain range relative to the first reference score as the first candidate group. Additionally, the device (200) can set a value corresponding to a certain percentage of the second reference score as the second reference value, and classify content types included within a certain range relative to the second reference score as the second candidate group.

[0187] According to another embodiment, the device (200) may dynamically set a first reference value and a second reference value by analyzing the matching score distribution for a plurality of artists or the matching score distribution for a plurality of content types.

[0188] For example, the device (200) may set a criterion to classify targets included within a certain range as a candidate group based on the average value, variance, or standard deviation of the matching scores. In this case, if the distribution of matching scores is concentrated in a specific interval, the range of the candidate group may be set relatively narrow, and if the distribution of matching scores is widely dispersed, the range of the candidate group may be set relatively wide.

[0189] In this way, the device (200) can set the first reference value and the second reference value in at least one of a fixed value, a ratio value, or a dynamic reference value based on a score distribution, so that the range of the candidate group can be flexibly adjusted according to brand characteristics, artist characteristics, or content type characteristics, and thereby a balance between the search range and calculation efficiency can be secured in the candidate combination generation process.

[0190] In step S508, the device (200) can generate multiple combinations by combining each artist classified as a first candidate group and each content type classified as a second candidate group. Here, the multiple combinations may include all possible combinations or some selected combinations generated by combining each artist included in the first candidate group and each content type included in the second candidate group.

[0191] For example, the device (200) can generate n × m combinations when the number of artists in the first candidate group is n and the number of content types in the second candidate group is m. Additionally, the device (200) may select only some combinations as valid combinations by additionally reflecting conditions such as contract availability, production period, budget range, or industry restrictions. That is, the device (200) can generate various combinations of artists and content types that have a similar suitability level for the first brand.

[0192] If the number of combinations of the first candidate group and the second candidate group exceeds a preset number of combinations, the device (200) may restrict the generation of combinations to using only the artists or content types with the highest matching scores.

[0193] In step S509, the device (200) can calculate a performance prediction score by predicting advertising performance with the first brand for each of the plurality of combinations. Here, the performance prediction score may be a value representing the expected advertising performance when a specific combination is used in actual advertising execution.

[0194] For example, the device (200) can predict the expected number of views, expected engagement rate, expected click-through rate, expected conversion rate, expected revenue contribution, or expected brand response rate for each combination by using the first brand information, information on each artist included in the combination, and information on each content type as input values, and can calculate a performance prediction score by synthesizing these. In addition, the device (200) may use a separate performance prediction artificial intelligence model for performance prediction by combination, and the performance prediction artificial intelligence model may be configured by learning past advertising performance data for each brand-artist-content combination. That is, the device (200) can quantify the potential for advertising execution performance for each of the multiple combinations and calculate a performance prediction score.

[0195] In step S510, the device (200) can identify the combination with the highest performance prediction score among the combinations as the first combination by comparing the performance prediction scores for each of the combinations. Here, the first combination may be the combination predicted to have the highest potential for advertising performance among the combinations of the first candidate group and the second candidate group.

[0196] For example, the device (200) may sort the performance prediction scores for each of the multiple combinations in descending order and identify the combination with the highest score as the first combination. Additionally, if there are combinations with the same score, the device (200) may determine the final first combination by additionally considering information such as artist contract feasibility, ease of content production, budget suitability, or brand preference. That is, the device (200) may identify the combination with the highest performance prediction score among the multiple combinations as the first combination.

[0197] In step S511, the device (200) can identify an artist included in the first combination as the second artist and identify a content type included in the first combination as the second type. Here, the second artist and the second type may refer to the components of the first combination identified as the top combination as a result of comparing multiple candidate combinations. That is, the device (200) can identify the artist and content type constituting the first combination selected as the optimal candidate combination as the second artist and the second type, respectively.

[0198] In step S512, the device (200) can identify the performance prediction score for the combination of the second artist and the second type as the second performance prediction score. Here, the second performance prediction score may be a score representing the expected advertising performance when the second artist and the second type included in the first combination are combined with the first brand.

[0199] For example, the device (200) can identify the score corresponding to the first combination among the performance prediction scores calculated for each of the plurality of combinations as the second performance prediction score. That is, the device (200) can identify the advertising performance potential of the optimal candidate combination as the second performance prediction score.

[0200] In step S601, the device (200) can determine whether the second performance prediction score is higher than the first performance prediction score. Here, the first performance prediction score is a reference performance prediction score for the combination of the first artist and the first type, which is the initial recommended combination, and the second performance prediction score may be a performance prediction score for the combination of the second artist and the second type derived through the candidate combination optimization process.

[0201] If it is confirmed that the second performance prediction score is higher than the first performance prediction score, in step S602, the device (200) can generate first matching information by matching the first brand, the second artist, and the second type of content. Here, the fact that the second performance prediction score is higher than the first performance prediction score may mean that a combination with a higher potential for advertising performance than the initial recommended combination has been discovered through candidate group expansion and combination comparison.

[0202] That is, the device (200) can generate first matching information by reflecting the second artist and second type when the candidate combination optimization result is superior to the reference combination.

[0203] If it is confirmed that the second performance prediction score is not higher than the first performance prediction score, in step S603, the device (200) can generate first matching information by matching the first brand, the first artist, and the first type of content. Here, the fact that the second performance prediction score is not higher than the first performance prediction score may mean that even if candidate group expansion and combination comparison were performed, no combination with a potential for performance exceeding the initial recommended combination was found. In this case, the device (200) can generate first matching information by maintaining the first artist and the first type, which are the initial recommended combinations, as they are.

[0204] That is, the device (200) can generate first matching information while maintaining the existing recommendation result if the candidate combination optimization result is not better than the reference combination.

[0205] According to one embodiment, the device (200) may collect performance results after the first matching information is used in actual advertising execution and use them for retraining candidate group composition criteria, performance prediction model, or final combination selection logic.

[0206] For example, if a combination of a second artist and a second type performs better than expected in an actual campaign, the device (200) can reflect the result as positive learning data to improve the accuracy of candidate group expansion and combination optimization. Conversely, if the actual performance is lower than the expected performance, the device (200) can use this as correction data to adjust the threshold setting method, prediction model parameters, or score reflection ratio. Through this, the device (200) can continuously improve the performance of selecting the final matching combination based on actual performance data accumulated during the process of repetitive advertising execution.

[0207] Accordingly, by expanding the candidate pool to include candidate artists and content types similar to the recommendation results, and predicting advertising performance for each of the multiple combinations to select the final combination, it becomes possible to derive the optimal combination that reflects combination synergy effects that are difficult to verify solely through recommendation results for individual elements. Furthermore, by comparing the initial recommended combination with the optimization results of the candidate combinations, both recommendation stability and optimization efficiency can be secured.

[0208] FIGS. 7 to 9 are flowcharts for explaining the process of displaying matching information according to one embodiment.

[0209] According to one embodiment, each step illustrated in FIGS. 7 to 9 can be performed in conjunction with step S211, and the device (200) can be configured to dynamically divide the screen area so that the matching result between the first brand, the artist, and the content type can be intuitively checked on the user terminal, and to select and display information to be displayed in each area according to priority.

[0210] Referring to FIGS. 7 through 9, first, in step S701, the device (200) may generate a first page for displaying first matching information and transmit it to a first user terminal (110). Here, the first matching information may include at least some of information regarding an artist and content type matched with the first brand and information regarding an artist and content type classified as a candidate group. Additionally, the first page may be a user interface screen for providing information regarding an artist and content type matched with the first brand to the user, and may be implemented in at least one form among a web page, an application screen, a dashboard screen, or other digital content display screen.

[0211] In step S702, the device (200) can identify the displayable area of ​​the first page as the first area. Here, the displayable area may refer to an area where content can actually be displayed, taking into account the screen resolution, screen ratio, and user interface layout of the first user terminal (110). For example, the displayable area may be defined as the actual content display area excluding the menu area at the top of the screen, the fixed button area at the bottom, or other system UI areas.

[0212] In step S703, the device (200) can determine the number of artists classified as a first candidate group as a first number. Here, the first candidate group may be a set of multiple artists whose matching score with the first brand is confirmed to be at a level similar to that of the reference artist.

[0213] In step S704, the device (200) can verify the number of content types classified as a second candidate group as a second number. Here, the second candidate group may be a set of multiple content types in which the matching score with the first brand is confirmed to be at a level similar to that of the reference content type.

[0214] In step S705, the device (200) can calculate a third value by adding the first value and the second value. Here, the third value may be a value representing the total number of candidate artists and candidate content types to be displayed on the first page.

[0215] In step S706, the device (200) can calculate a first ratio by dividing a first value by a third value, and calculate a second ratio by dividing a second value by a third value. Here, the first ratio may mean the proportion of the artist candidate group in the total candidate set, and the second ratio may mean the proportion of the content type candidate group in the total candidate set.

[0216] In step S707, the device (200) may calculate a second area by multiplying a first area by a first ratio, and a third area by multiplying a first area by a second ratio. Here, the second area may refer to the size of an area for displaying artist-related information, and the third area may refer to the size of an area for displaying content type-related information.

[0217] In step S708, the device (200) may set a first area with a size of a second area within a first page and set a second area with a size of a third area within a first page. At this time, the first area may be set as an area for displaying artist-related information, and the second area may be set as an area for displaying content type-related information.

[0218] In step S801, the device (200) can identify the minimum display area required to display detailed information of an artist as the fourth area. Here, the detailed information of an artist is information intended to provide in-depth information about a specific artist and may include at least one of the artist's name, profile image, representative content, activity platform, fan base characteristics, history of participation in past advertising campaigns, advertising performance indicators, or major activity fields. At this time, the fan base characteristics may include the age distribution, gender distribution, interest distribution, or activity platform information of the user group that primarily consumes the artist, and the advertising performance indicators may include views, engagement rate, click-through rate, conversion rate, or brand response rate increase rate.

[0219] In step S802, the device (200) can calculate the fifth area by subtracting the fourth area from the second area.

[0220] In step S803, the device (200) may set a 1-1 area with a size of a fourth area within a first area, and a 1-2 area with a size of a fifth area within a first area. Here, the 1-1 area may be used as an area for displaying detailed information of the final matching artist, and the 1-2 area may be used as an area for displaying a list of candidate artists.

[0221] In step S804, the device (200) can identify the artist matched with the first brand as the final matching artist based on the first matching information.

[0222] For example, if the first matching information is generated through step S602, the device (200) can identify the second artist matched with the first brand as the final matching artist, and if the first matching information is generated through step S603, the first artist matched with the first brand as the final matching artist.

[0223] In step S805, the device (200) may generate a first list including a first artist and artists classified as a first candidate group. Here, the first list may mean a set of artist candidates including a first artist and multiple artists classified as a first candidate group, and may be used as an initial candidate list including artists that can be matched with a first brand.

[0224] In step S806, the device (200) may generate a second list containing artists excluding the final matching artist from the first list. Here, the second list may be a list containing artists excluding the final matching artist from the first list, and may be used as a comparison target list containing the remaining candidate artists excluding the final matching artist.

[0225] In step S807, the device (200) can give higher priority to each artist included in the second list as the higher the matching score with the first brand.

[0226] In step S808, the device (200) can identify the minimum display area required to display the artist's basic information as the sixth area. Here, the artist's basic information is a minimum level of information for identifying the artist and may include at least one of the artist's name, profile image, activity platform information, or main content field. Additionally, the artist's basic information may be used as information for briefly displaying the artist list.

[0227] In step S809, the device (200) may set a first rank using the quotient portion of the value obtained by dividing the fifth area by the sixth area. Here, the first rank may mean the maximum number of candidate artists that can be displayed simultaneously in the first-second area.

[0228] In step S810, the device (200) may generate a third list containing artists whose priority is within the first rank in the second list. Here, the third list may be a list containing artists whose priority based on matching scores is within the first rank among the artists included in the second list, and may mean a set of candidate artists to be actually displayed on the first page.

[0229] In step S901, the device (200) can identify the minimum display area required to display detailed information of the content type as the seventh area. Here, the detailed information of the content type is information for describing the characteristics of a specific content type and may include at least one of the content type name, content presentation method, characteristics of the main audience, content advertising performance history, major publishing platforms, or content response data. At this time, the content presentation method may refer to a content production format such as information-delivery content, experience-oriented content, story-type content, challenge-type content, or interview-type content.

[0230] In step S902, the device (200) can calculate the eighth area by subtracting the seventh area from the third area.

[0231] In step S903, the device (200) may set a 2-1 area with the size of a 7 area within a 2 area, and a 2-2 area with the size of an 8 area within a 2 area. Here, the 2-1 area may be used as an area for displaying detailed information of the final matching content type, and the 2-2 area may be used as an area for displaying a list of candidate content types.

[0232] In step S904, the device (200) can identify the content type matched with the first brand as the final matching content type based on the first matching information.

[0233] For example, if the first matching information is generated through step S602, the device (200) can identify the second type matched with the first brand as the final matching content type, and if the first matching information is generated through step S603, the first type matched with the first brand as the final matching content type.

[0234] In step S905, the device (200) may generate a fourth list including content types classified into a first type and a second candidate group. Here, the fourth list may mean a set of content type candidates including multiple content types classified into a first type and a second candidate group, and may be used as an initial candidate list including content types that can be matched with the first brand.

[0235] In step S906, the device (200) may generate a fifth list containing content types excluding the final matching content type from the fourth list. Here, the fifth list may be a list containing content types excluding the final matching content type from the fourth list, and may be used as a comparison target list containing the remaining candidate content types excluding the final matching content type.

[0236] In step S907, the device (200) can assign a higher priority to each content type included in the fifth list as the higher the matching score with the first brand.

[0237] In step S908, the device (200) can identify the minimum display area required to display basic information of the content type as the ninth area. Here, the basic information of the content type is a minimum level of information for briefly identifying the content type, and may include at least one of a content type name, a representative icon, a major platform, or a content category.

[0238] In step S909, the device (200) may set a second rank using the quotient portion of the value obtained by dividing the eighth area by the ninth area. Here, the second rank may mean the maximum number of candidate content types that can be simultaneously displayed in the second-2 area.

[0239] In step S910, the device (200) may generate a sixth list containing content types that have a priority within the second rank in the fifth list. Here, the sixth list may be a list containing content types that have a priority within the second rank based on a matching score among the content types included in the fifth list, and may mean a set of candidate content types to be actually displayed on the first page.

[0240] In step S911, the device (200) can control the display of detailed information of the final matching artist on the first page in the first-1 area, control the display of the third list in the first-2 area, control the display of detailed information of the final matching content type in the second-1 area, and control the display of the sixth list in the second-2 area.

[0241] Accordingly, information on artists and content types matched with the brand can be efficiently displayed on the user terminal, and by providing candidate artists and candidate content types together, users can intuitively check the matching results while comparing various alternatives. Additionally, by dynamically dividing the screen area according to the number of candidates and automatically adjusting the number of displayable items, the matching results can be effectively provided while maintaining information readability in various screen environments.

[0242] FIG. 10 is an example diagram of the configuration of a device according to one embodiment.

[0243] A device (200) according to one embodiment includes a processor (210) and a memory (220). The processor (210) may include at least one device described with reference to FIGS. 1 through 9 or may perform at least one method described with reference to FIGS. 1 through 9. A person or organization using the device (200) may provide services related to some or all of the methods described with reference to FIGS. 1 through 9.

[0244] The memory (220) may store information related to the methods described above or store a program in which the methods described below are implemented. The memory (220) may be volatile memory or non-volatile memory.

[0245] The processor (210) can execute a program and control the device (200). The code of the program executed by the processor (210) can be stored in memory (220). The device (200) is connected to an external device (e.g., a personal computer or a network) through an input / output device (not shown in the drawing) and can exchange data through wired or wireless communication.

[0246] The device (200) can be used to train an artificial intelligence model or to use a trained artificial intelligence model. The memory (220) may contain an artificial intelligence model that is being trained or has been trained. The processor (210) can train or execute an artificial intelligence model algorithm stored in the memory (220). The training device for training the artificial intelligence model and the device (200) for using the trained artificial intelligence model may be the same or separate.

[0247] The embodiments described above may be implemented as hardware components, software components, and / or combinations of hardware and software components. For example, the devices, methods, and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0248] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0249] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or command the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0250] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based on the above. For example, suitable results may be achieved even if the described techniques are performed in a different order than described, and / or if the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0251] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.

Claims

Claim 1 A method for matching between an AI-based brand, an artist, and content, performed by a device, comprising: receiving a matching request for an artist and content for a first brand from a first user terminal; obtaining first brand information, which is information related to the first brand; encoding the first brand information to generate a first input signal; inputting the first input signal to a first AI model trained to select a recommended artist through the analysis of brand information; when the recommended artist of the first brand is selected as the first artist through the first input signal, obtaining a first output signal representing the first artist from the first AI model; determining the recommended artist of the first brand as the first artist based on the first output signal; inputting the first input signal to a second AI model trained to select a recommended content type through the analysis of brand information; when the recommended content type of the first brand is selected as the first type through the first input signal, obtaining a second output signal representing the first type from the second AI model; and determining the recommended content type of the first brand as the first type based on the second output signal. The method comprises: a step of determining; a step of matching the first brand, the first artist, and the first type of content to generate first matching information; and a step of transmitting the first matching information to the first user terminal, wherein the first artificial intelligence model identifies the first brand information based on the first input signal, identifies the brand image, target customer base, and brand campaign performance of the first brand based on the first brand information, selects the first artist, which is one of a plurality of pre-registered artists, as the subject of analysis, obtains first artist information which is information related to the first artist, and, based on the first artist information, identifies the artist image, fan base, andThe artist advertising performance is verified, the similarity between the brand image of the first brand and the artist image of the first artist is calculated as the first similarity, the similarity between the target customer base of the first brand and the fan base of the first artist is calculated as the second similarity, the similarity between the brand campaign performance of the first brand and the artist advertising performance of the first artist is calculated as the third similarity, and the first similarity, the second similarity, and the third similarity are summed to calculate a matching score between the first brand and the first artist. When matching scores between the first brand and the plurality of artists are each calculated, if the matching scores between the first brand and the plurality of artists are compared and it is confirmed that the matching score between the first brand and the first artist is the highest, the recommended artist of the first brand is selected as the first artist. The second artificial intelligence model verifies the first brand information based on the first input signal, verifies the brand image, target customer base, and brand campaign performance of the first brand based on the first brand information, and analyzes a first type, which is one of a plurality of pre-registered content types. Selecting a target, obtaining information on a first content type related to the first type, and based on the first content type information, verifying the content image, the main audience of the content, and the advertising performance of the first type, calculating the similarity between the brand image of the first brand and the content image of the first type as a fourth similarity, calculating the similarity between the target audience of the first brand and the main audience of the first type as a fifth similarity, calculating the similarity between the brand campaign performance of the first brand and the advertising performance of the first type as a sixth similarity, and summing the fourth similarity, the fifth similarity, and the sixth similarity to calculate a matching score between the first brand and the first type.When matching scores are calculated for each of the first brand and the plurality of content types, if the matching score between the first brand and the plurality of content types is compared and it is confirmed that the matching score between the first brand and the first type is the highest, the recommended content type of the first brand is selected as the first type, and the step of generating the first matching information comprises: a step of predicting advertising performance with the first brand for a combination of the first artist and the first type to calculate a first performance prediction score; a step of verifying the matching score with the first brand for each of the plurality of artists; a step of setting the matching score between the first brand and the first artist as a first reference score; a step of classifying among the plurality of artists artists as a first candidate group for which the difference between each artist's matching score and the first reference score is less than or equal to a preset first reference value; a step of verifying the matching score with the first brand for each of the plurality of content types; a step of setting the matching score between the first brand and the first type as a second reference score; and among the plurality of content types, content types for which the difference between each content type's matching score and the second reference score is less than or equal to a preset second reference value. A step of classifying into a second candidate group; a step of generating multiple combinations by combining each artist classified into the first candidate group and each content type classified into the second candidate group; a step of predicting advertising performance with the first brand for each of the multiple combinations and calculating a performance prediction score; a step of comparing the performance prediction scores for each of the multiple combinations and identifying the combination with the highest performance prediction score among the multiple combinations as the first combination; identifying the artist included in the first combination as the second artist, andA method for matching between an AI-based brand, an artist, and content, comprising: a step of identifying a content type included in the first combination as a second type; a step of identifying a performance prediction score for the combination of the second artist and the second type as a second performance prediction score; a step of generating first matching information by matching the first brand, the second artist, and the content of the second type if it is confirmed that the second performance prediction score is higher than the first performance prediction score; and a step of generating first matching information by matching the first brand, the first artist, and the content of the first type if it is confirmed that the second performance prediction score is not higher than the first performance prediction score. Claim 2 delete Claim 3 delete

Images