Live commerce provision system using AI show host

Abstract

The present invention relates to a live commerce provision system, and more specifically, to a live commerce provision system using an AI show host that enables live commerce targeting various countries overseas by overcoming linguistic and cultural constraints, by translating a broadcast scenario introducing a product into the languages ​​of various countries using a multilingual translation means, and then executing live commerce in which an AI show host created considering the cultural characteristics of each country appears with product images to promote the product, and by enabling real-time communication through AI learning based on LLM (Large Language Model) and then outputting the AI ​​show host's response comments in response to real-time chat input by viewers during the live commerce, thereby enabling live commerce.

Description

Technology Field

[0001] The present invention relates to a live commerce provision system, and more specifically, to a live commerce provision system using an AI show host that enables live commerce targeting various countries overseas by overcoming linguistic and cultural constraints, by translating a broadcast scenario introducing a product into the languages ​​of various countries using a multilingual translation means, and then executing live commerce in which an AI show host created considering the cultural characteristics of each country appears with product images to promote the product, and by enabling real-time communication through AI learning based on LLM (Large Language Model) and then outputting the AI ​​show host's response comments in response to real-time chat input by viewers during the live commerce, thereby enabling live commerce. Background Technology

[0002] Recently, the live commerce market has been growing rapidly not only in Korea but also in countries around the world, including China and Southeast Asia, and global platforms such as Shopee are also supporting live commerce services.

[0003] Live commerce is an e-commerce model that introduces and sells products to viewers through real-time video streaming. Consumers access product information while watching internet broadcasts, similar to home shopping, and purchases are made through real-time interaction between the show host introducing the products and the viewers wishing to buy, much like in an offline store.

[0004] Unlike conventional home shopping, where product information is unilaterally transmitted via television, this type of live commerce differs in that active, two-way exchange of information related to product purchases can take place between show hosts and viewers.

[0005] Consequently, while live commerce has the advantage of allowing experts from various fields, such as product representatives or influencers rather than professional broadcasters, to appear and freely exchange opinions with viewers, the capabilities of the show host become even more crucial for natural and spontaneous communication with viewers, in addition to product introductions.

[0006] In particular, to execute global live commerce, it is important to recruit local show hosts who are proficient in the local culture and language, possess the ability to convey information for product introductions, and have the quick wit to communicate with viewers; however, this was by no means an easy task for foreign companies unfamiliar with local conditions.

[0007] Accordingly, a technology for conducting real-time broadcasting using an AI human model generated using AI technology has recently been introduced, as disclosed in Korean Public Patent No. 10-2023-0120899.

[0008] However, in the case of such conventional 'method and system for providing live commerce through an AI human model,' although an AI human model is created to conduct the broadcast, the AI ​​human model conducts sales based on a specific script prepared in advance according to the product. Consequently, there is a problem in that it is somewhat distant from the interaction with viewers shown by actual show hosts during live commerce broadcasts, and is operated more like home shopping where information is unilaterally delivered.

[0009] Accordingly, the need for live commerce that enables active communication with viewers—as if conducted by a real show host—while utilizing AI technology, and can be easily applied not only domestically but also to live commerce broadcasts in various countries overseas, is increasing even further. The problem to be solved

[0010] The present invention aims to solve the above-mentioned problems by providing a live commerce provision system using an AI show host that enables live commerce targeting various countries overseas, free from linguistic and cultural constraints, by translating a broadcast scenario introducing a product into the languages ​​of various countries using a multilingual translation means, executing live commerce in which an AI show host created considering the cultural characteristics of each country appears with product images to promote the product, and enabling real-time communication by generating response comments from the AI ​​show host that respond to real-time chat input by viewers during the live commerce through AI learning based on an LLM (Large Language Model) and outputting them as voice and video.

[0011] The above and other objects and advantages of the present invention will become apparent from the following description describing preferred embodiments. means of solving the problem

[0012] A live commerce providing system using an AI show host according to an embodiment of the present invention for achieving the above-mentioned purpose may include: an AI show host generating image of an AI show host who is the main subject of the live commerce broadcast and an image of a product to be sold, generating a broadcast scenario to be conducted by the AI ​​show host for product sales, and then generating broadcast content in which location information of the AI ​​show host moving on the broadcast screen and location information of the product image are set according to the broadcast scenario and storing it in a database; and a live commerce operating server that specifies a country to be broadcast live commerce, translates the broadcast scenario into the language of the country, corrects the mouth shape and facial expression or gesture of the AI ​​show host according to the translated language, conducts live commerce by transmitting the corrected broadcast content over a communication network, and executes two-way communication with the viewer by modifying and displaying the AI ​​show host image and the product image to additionally execute response comments or actions in response to real-time chat input by the viewer during the live commerce.

[0013] At this time, the AI ​​show host generation server may include: an AI show host generation unit that generates an avatar having a human-like appearance, wherein the face of the avatar is generated to have multiple facial expressions capable of expressing various emotions, and the hands of the avatar are generated to have multiple hand movements capable of taking various gestures, including the action of grasping a product, and stores them in a database; a product image generation unit that generates products to be sold through live commerce by 3D scanning or stores various product images received from a manufacturer in a database; and a broadcast content generation unit that generates a broadcast scenario in which the AI ​​show host and the product image are displayed together on a broadcast screen to provide explanations and promotions about the product to viewers, specifies the facial expressions and gestures taken by the AI ​​show host that are displayed on the screen during live commerce according to the broadcast scenario, and sets location information of the product image displayed together with the AI ​​show host to generate broadcast content.

[0014] In addition, the AI ​​show host generation unit can generate an AI show host with an appearance that is familiar or preferred in each country by creating a prototype show host with a basic human appearance as an avatar and then adjusting the show host's clothing, skin color, and facial shape according to the country or region where the live commerce is to be broadcast.

[0015] In addition, the broadcast content generation unit can create a realistic screen where the AI ​​show host explains the product while holding and moving the product directly by matching the location information where the AI ​​show host's hand is exposed on the screen with the location information where the product image is exposed, thereby displaying them at the same location.

[0016] In addition, the broadcast content generation unit can designate the area between the position information of the two hands of the AI ​​show host as the area where the product image is placed, set the position information of the product image inside the area, and then display the product image while rotating it 360° within the space inside the area.

[0017] Additionally, the live commerce operation server may include: a broadcast content correction unit that specifies the country to which the live commerce broadcast is to be transmitted, translates the broadcast scenario into the language of the country, and corrects the clothing, mouth shape, facial expression, voice tone, or gestures of the AI ​​show host according to the translated language to generate broadcast content for the country; a live commerce execution unit that transmits the broadcast content for the country over a communication network so that the AI ​​show host promotes products in the language of each country and induces purchase; a viewer identification unit that identifies a viewer using the ID of a viewer who inputs real-time chat during the live commerce, and determines whether the identified viewer has a history of participating in past live commerce and whether they have a history of purchasing products through live commerce; and a viewer communication unit that generates feedback content using the movements and voice of the AI ​​show host to provide product information or purchase-related information desired by the viewer based on the real-time chat input by the identified viewer, and then transmits it through live commerce.

[0018] In addition, the viewer communication unit can conduct real-time communication based on LLM using comments entered by the viewer, generate product information or purchase-related information desired by the viewer, or response comments that can induce a purchase, and generate and play the voice, mouth shape, or gesture of the AI ​​show host to output the generated response comments, thereby enabling real-time two-way communication with the viewer.

[0019] In addition, the viewer communication unit can perform two-way communication by using a dataset of various types of chats entered by viewers in past live commerce and response comments that induce viewers to purchase products, learning AI based on LLM (Large Language Model), and using the results of such learning to generate and output response comments for comments registered during the current live commerce as feedback.

[0020] In addition, the viewer communication unit can perform individual responses to viewers by adjusting the facial expression or voice tone of the AI ​​show host according to the gender of the viewer, generating response comments with a friendly tone that allows for emotional communication through empathy when the viewer is female, and generating response comments that explain technical content with a polite tone when the viewer is male.

[0021] In addition, the viewer communication unit can generate response comments for viewers currently watching the live commerce based on the usage experiences of buyers who purchased the same product during a previous live commerce session stored in a database, thereby enabling empathy in sharing the product's usage experience. Effects of the invention

[0022] According to an embodiment of the present invention, a broadcast scenario introducing a product is translated into the languages ​​of various countries by a multilingual translation means, and an AI show host is corrected and generated considering the cultural characteristics of each country to execute live commerce. Therefore, products can be sold domestically while communicating with viewers in various countries without the need to recruit local show hosts from overseas, thereby promoting the entry of domestic companies into overseas markets.

[0023] In addition, the present invention enables real-time communication by generating response comments from an AI show host that respond to real-time chat input by viewers during live commerce through AI learning based on an LLM (Large Language Model) and outputting them as voice and video, thereby moving beyond the one-way provision of product information when there is no professional show host and maintaining the utility of live commerce that allows for two-way real-time communication between the AI ​​show host and the viewers.

[0024] This invention has the effect of expanding the accessibility to overseas sales for small and medium-sized enterprises that have been hesitant to engage in overseas online sales due to difficulties such as language barriers and securing show hosts for product promotion. Brief explanation of the drawing

[0025] FIG. 1 is an overall system diagram of a live commerce provision system using an AI show host according to the present invention. FIG. 2 is a conceptual diagram showing live commerce being executed in the language of each country according to the present invention. FIG. 3 is a block diagram of a live commerce provision system using an AI show host according to the present invention. FIG. 4 is an example diagram illustrating the generation and provision of queries and responses to product data based on RAG according to the present invention. FIG. 5 is an exemplary diagram showing the flow of generating an AI show host avatar and voice content according to the present invention. FIG. 6 is an exemplary diagram illustrating the advancement of an LLM system for live commerce conducted by an AI show host according to the present invention. FIG. 7 is a flowchart showing live commerce conducted using an AI show host according to the present invention. Specific details for implementing the invention

[0026] The present invention will be described in detail below with reference to the embodiments and drawings. These embodiments are presented merely as examples to explain the invention more specifically, and it will be obvious to those skilled in the art that the scope of the invention is not limited by these embodiments.

[0027] Furthermore, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention pertains, and in the event of a conflict, the description in this specification, including the definitions, shall prevail.

[0028] In order to clearly explain the proposed invention in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification have been given similar reference numerals. Furthermore, when a part is described as "comprising" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0029] Hereinafter, an embodiment of a live commerce provision system using an AI show host according to the present invention will be described in detail with reference to FIGS. 1 to 7.

[0030] FIG. 1 is an overall system diagram of a live commerce provision system using an AI show host according to the present invention, FIG. 2 is a conceptual diagram showing live commerce being executed in the language of each country according to the present invention, and FIG. 3 is a block diagram of a live commerce provision system using an AI show host according to the present invention.

[0031] Referring to FIGS. 1 to 3, a live commerce providing system using an AI show host according to an embodiment of the present invention may include: an AI show host generating server (100) that generates an image of an AI show host who is the main subject of the live commerce broadcast and an image of a product to be sold, generates a broadcast scenario to be conducted by the AI ​​show host for product sales, and then generates broadcast content in which location information of the AI ​​show host moving on the broadcast screen and location information of the product image are set according to the broadcast scenario and stores it in a database; and a live commerce operating server (200) that specifies a country to be broadcast the live commerce broadcast, translates the broadcast scenario into the language of the country, corrects the mouth shape and facial expression or gesture of the AI ​​show host according to the translated language, transmits the corrected broadcast content over a communication network to conduct live commerce, and performs two-way communication with the viewer by modifying and displaying the AI ​​show host image and the product image to additionally execute response comments or actions in response to real-time chat input by the viewer during the live commerce.

[0032] The above AI show host generation server (100) may include an AI show host generation unit (110) that generates an avatar having a human-like appearance, wherein the face of the avatar is generated to have multiple expressions capable of expressing various emotions, and the hands of the avatar are generated to have multiple hand movements capable of taking various gestures, including the action of holding a product, and stores them in a database; a product image generation unit (120) that generates a product to be sold through live commerce by 3D scanning it or stores various product images received from a manufacturer in a database; and a broadcast content generation unit (130) that generates a broadcast scenario in which the AI ​​show host and the product image are displayed together on a broadcast screen while explaining and promoting the product to viewers, specifies the expressions and gestures taken by the AI ​​show host that are displayed on the screen during live commerce according to the broadcast scenario, and sets the location information of the product image that is displayed together with the AI ​​show host to generate broadcast content.

[0033] The above AI show host generation unit (110) generates a show host who will conduct a broadcast while introducing products and communicating with viewers during live commerce as an avatar with a human-like appearance based on AI technology and stores it in a database (not shown).

[0034] To this end, the AI ​​show host generation unit (110) can generate an AI show host with an appearance that is familiar or preferred in each country by generating a prototype show host with a basic human appearance as an avatar, and then adjusting the show host's clothing, skin color, face shape, etc. according to the country (e.g., Thailand or Vietnam, etc.) or region where live commerce will be broadcast.

[0035] The clothing, skin color, and hairstyle of the AI ​​show host generated in this manner can be varied depending on changes in the product to be sold or the season, and it is also possible to generate multiple AI show hosts with various appearances and store them in a database in advance.

[0036] At this time, the AI ​​show host generation unit (110) can generate the prototype show host based on a graphic having a human appearance and generate the AI ​​show host based thereon, and can also increase the realism or naturalness of the AI ​​show host's movements, facial expressions, or gestures by first filming a video of a real model promoting a product according to a broadcast scenario to generate the prototype show host and its movements, and generating the AI ​​show host based thereon.

[0037] In addition, the AI ​​show host generation unit (110) can generate various facial expressions by changing the mouth shape, eye size, or facial muscle parts of the prototype show host so that various emotional expressions such as joy, pleasure, welcome, surprise, and regret can be visually recognized through facial changes, and then store them in a database.

[0038] In addition, the broadcast content generation unit (130) can make the AI ​​show host display a variety of facial expressions that match the broadcast content during live commerce by pre-setting the AI ​​show host's facial expressions to change and be displayed according to the broadcast scenario, thereby enabling a more realistic broadcast.

[0039] In addition, the AI ​​show host generation unit (110) can generate hand movements to make the prototype show host's hand grasp a product, or generate multiple hand movements that take various gestures commonly used according to the local culture of each country during product introduction.

[0040] In this way, so that the AI ​​show host can display human-like facial expressions or hand gestures during live commerce, the AI ​​show host generation unit enables a more realistic and dynamic live commerce environment through motion graphics generated by reflecting local culture.

[0041] In addition, the product image generation unit (120) may generate a product image by 3D scanning the product when registering a product for sale commissioned by a seller, or receive promotional materials regarding the exterior, interior, or components of the product from a manufacturer as a product image and store it in a database.

[0042] At this time, the product image generation unit (120) can generate not only a simple flat image but also a full 3D image that can be exposed when the AI ​​show host holds and rotates the product, since the AI ​​show host can show the entire exterior while rotating the product.

[0043] In addition, the broadcast content generation unit (130) generates a broadcast scenario that combines an AI show host image and a product image displayed on a broadcast screen to promote products sold by the AI ​​show host, and generates broadcast content that induces product sales while communicating in real-time with viewers participating in live commerce, and stores it in a database.

[0044] At this time, the broadcast content generation unit (130) can, of course, generate the voice of an AI show host according to a broadcast scenario using text-to-speech (TTS) and store it together.

[0045] The above broadcast content generation unit (130) generates broadcast content by specifying the facial expressions and gestures of the AI ​​show host that are displayed on the screen according to the broadcast scenario, and specifying the location information of the product images that are displayed together.

[0046] In addition, the broadcast content generation unit (130) can create a realistic screen where the AI ​​show host explains the product while holding and moving the product, just like in live commerce conducted by a real show host, by matching the location information where the AI ​​show host's hand is exposed on the screen with the location information where the product image is exposed, so that they are displayed at the same location.

[0047] Additionally, the broadcast content generation unit (130) may be configured to designate the area between the position information of the two hands as an area where a product image is placed, set the position information of the product image inside the area, and then display the product image while rotating it 360° within the space inside the area.

[0048] In this way, when the above product image is displayed in the area between the two hands of the AI ​​show host, not only can the product image be avoided from being obscured by the hand image, but the product image to be displayed can also be diversified, such as by enabling separate display of the product image or display of its internal composition.

[0049] Additionally, the live commerce operation server (200) may include a broadcast content correction unit (210) that specifies the country to which the live commerce broadcast will be transmitted, translates the broadcast scenario into the language of the country, and corrects the clothing, mouth shape, facial expression, voice tone, or gestures of the AI ​​show host according to the translated language to generate broadcast content for the country; a live commerce execution unit (220) that transmits the broadcast content for the country over a communication network so that the AI ​​show host promotes products in the language of each country and induces purchase; a viewer identification unit (230) that identifies the viewer using the ID of the viewer who inputs real-time chat during the live commerce, determines whether the identified viewer has a history of participating in past live commerce and whether they have a history of purchasing products through live commerce; and a viewer communication unit (240) that generates feedback content using the movements and voice of the AI ​​show host to provide product information or purchase-related information desired by the viewer based on the real-time chat input by the identified viewer, and then transmits it through live commerce.

[0050] The above broadcast content correction unit (210) can generate broadcast content specialized for that country by translating the above broadcast scenario into the language of the country where the live commerce broadcast will be transmitted, correcting the appearance of the AI ​​show host, such as clothing and skin tone, by reflecting the cultural characteristics of the country, and correcting or adding the mouth shape, facial expression, voice tone, or gesture of the AI ​​show host according to the pronunciation of the translated language.

[0051] At this time, the broadcast content correction unit (210) can understand the content of the broadcast scenario and perform translation into the language of each country based on a Large Language Model (LM) that learns vast amounts of text data and understands the natural language of each country, thereby enabling a more friendly explanation of the product and a natural approach to the viewer.

[0052] In addition, the broadcast content correction unit (210) can create broadcast content for the country by correcting the mouth shape of the AI ​​show host or adding gestures while taking into account the pronunciation of the translated broadcast scenario, thereby preventing viewers watching the live commerce broadcast from feeling a sense of unfamiliarity due to the discrepancy between the voice and the mouth shape.

[0053] In this way, the broadcast content correction unit (210) automatically translates the content into the language of the country where live commerce is to be executed and corrects the AI ​​show host, thereby enabling small and medium-sized enterprises, which previously had difficulty entering the global market due to a lack of capital or technology, to conduct product introduction and sales through live commerce in multiple countries including China, Southeast Asia, and South America.

[0054] The above live commerce execution unit (220) matches the voice translated by the above broadcast content correction unit (210) and transmits the broadcast content for the corresponding country, which has been corrected by the AI ​​show host, through a pre-configured global platform such as Shopee. Accordingly, it is possible to promote and sell products while communicating with viewers in various countries overseas from within the country without having to hire a local show host from abroad.

[0055] The above viewer identification unit (230) can identify a viewer by using the ID of the viewer who entered the comment when a comment is entered for inquiries about a product, etc., while live commerce is being conducted while playing broadcast content for the country concerned.

[0056] Afterwards, the viewer identification unit (230) can search for the same ID in the past live commerce progress records stored in the database to determine whether the identified viewer has a history of participating in past live commerce and whether they have a history of purchasing products.

[0057] In addition, the viewer communication unit (240) can perform real-time communication based on LLM using comments entered by the viewer, generate product information or purchase-related information or response comments that can induce purchase that the viewer wants, generate the voice and mouth shape or gesture of the AI ​​show host to output the generated response comments, and play them while live commerce is in progress, thereby enabling real-time two-way communication with the viewer.

[0058] In the case of typical live commerce hosted by a show host, the host promoting the product frequently checks the comment section; when a comment related to the product currently being promoted is found, the host relies on their understanding of the product and memory to provide information through the broadcast to resolve the viewers' questions.

[0059] At this time, the show host strengthened the bond with viewers by expressing familiarity before delivering product information, such as by looking at the ID of a viewer who posted a comment and greeting the viewer with a welcome if it was a familiar ID or by asking about their satisfaction with a product purchased through past live commerce. In this case, the show host would call out the ID during the broadcast and greet the viewer with a welcoming expression and tone of voice.

[0060] However, in the case of AI show hosts that unilaterally provide product descriptions according to a fixed broadcast script, it becomes difficult to build a bond with viewers like a human by utilizing comments registered during the broadcast.

[0061] Previously, live commerce was also conducted using avatars such as AI human models, but in this case, the AI ​​human model only unilaterally provided information about the product, and communication with viewers who entered comments had to be done in the form of responding comments after a separate person checked the comment section. As a result, not only was the advantage of two-way communication in live commerce not utilized, but training and managing operational personnel was difficult, and language limitations could also be encountered.

[0062] In contrast, the above-mentioned viewer communication unit (240) can identify the viewer using the ID that registers the comment, identify past communication content or purchase history from the database, and then proceed with two-way communication by recognizing the viewer and greeting them or expressing familiarity, just as a general show host does.

[0063] To this end, the viewer communication unit (240) learns AI (artificial intelligence) based on LLM (Large Language Model) using a dataset of various types of chats entered by viewers in past live commerce and response comments that induce conversation to allow viewers to purchase products, and generates and outputs appropriate response comments for comments registered during the current live commerce as feedback according to the results of this learning, thereby enabling two-way communication.

[0064] Unlike conventional e-commerce, where consumers make purchasing decisions on their own after individually searching for product information and reading reviews or descriptions, or conventional home shopping, where consumers make purchasing decisions alone after watching product descriptions or demonstration videos based on a set scenario, live commerce facilitates consumers' purchasing decisions by allowing them to exchange questions and answers about the product through real-time chat with the show host demonstrating the product.

[0065] Accordingly, show hosts in live commerce go beyond simply introducing products and play a key role in inducing purchasing decisions through real-time communication with viewers.

[0066] Consequently, as a show host's professionalism, charm, and communication skills serve as important factors in building consumer trust and loyalty, most show hosts possess response strategies to quickly handle questions arising from real-time chat in live commerce.

[0067] Recent research has shown that the purchase conversion rate is highest when the response time to real-time chat during live commerce is within 5 seconds, and decreases sharply when it takes 10 seconds or more. In other words, if the response time after a consumer asks a question is prolonged, the likelihood of abandoning live commerce increases, and the consumer is more likely to seek information on their own through other platforms or search engines.

[0068] Accordingly, the viewer communication unit (240) can be configured to prepare and store standard answers in advance for frequently asked questions (e.g., delivery, discount, or stock, etc.) so that a quick response is possible, and then analyze the chat content entered by the viewer to find a suitable standard answer so that the AI ​​show host can respond quickly.

[0069] At this time, if inquiries about the same content are repeated by a large number of viewers and the AI ​​show host repeats the same response comments in response, other viewers may easily feel fatigued. Therefore, for some inquiries related to delivery, discounts, or payment methods, the AI ​​chatbot may be configured to automatically respond in the form of comments to the inquiries.

[0070] Furthermore, in live commerce, a show host's communication style is understood to be directly linked to the formation of consumer trust. In particular, female consumers tend to place greater importance on a show host's interpersonal communication skills, while male consumers tend to place more trust in the show host's expertise. Consequently, when the viewers participating in the chat are female, a friendly and attentive response is required, whereas when the viewers are male, a response emphasizing professionalism may be necessary.

[0071] Accordingly, the viewer communication unit (240) can be configured to enable individual responses to viewers by correcting the facial expressions or voice tone of the AI ​​show host according to gender when the gender of a specific viewer can be identified.

[0072] That is, the above viewer communication unit (240) can be configured to provide a customized response to the individual viewer by generating a response message with content that allows for emotional communication through empathy in a friendly tone when the viewer is female, and generating a response message with technical content in a polite tone when the viewer is male.

[0073] In addition, the viewer communication unit (240) may improve trust and friendliness by allowing the AI ​​show host to call out the name or ID of the viewer being responded to when responding to real-time chat, thereby making the viewer aware that the current response message is a personalized response for that viewer.

[0074] While many show hosts recommend purchasing products by sharing their personal experiences to connect with viewers, AI show hosts find it difficult to share that kind of empathy because they lack such personal experience.

[0075] Accordingly, the viewer communication unit (240) can enable empathy in sharing experiences by generating response comments for viewers currently watching live commerce based on the user experiences of buyers who purchased the same product during a previous live commerce session stored in a database. At this time, it goes without saying that the reliability of the user experience can be improved by clearly disclosing the buyers who made the purchase during the live commerce session held at a specific time and place.

[0076] Additionally, the viewer communication unit (240) may be configured to generate an immediate response message regarding purchase-related information, such as the price, discount, or promotion of a product that is frequently inquired about by viewers, and notify the viewer accordingly.

[0077] Accordingly, when there is an inquiry from a viewer regarding the receipt of a free gift upon purchasing a product, the system can be configured to respond by generating a response message that induces a quick purchase decision by emphasizing the urgency of time or the scarcity of remaining stock, such as, "If you purchase during the broadcast now, you will receive an additional free gift. Also, if you purchase within 30 minutes, you will receive additional benefits!"

[0078] In addition, since such purchase-related information is determined before the live commerce broadcast, response messages regarding the passage of time or the decrease in remaining quantity can be generated in advance, and when that time or remaining quantity reaches that level, they can be played as messages to induce a purchase decision.

[0079] In addition, the viewer communication unit (240) may generate a response message that can restore trust by promising empathy and improvement in response to negative feedback or claims from viewers that are posted in real-time chat.

[0080] When the above-mentioned viewer communication unit (240) receives a chat from a viewer that may include various contents such as product-related information, purchase-related information, or negative feedback, it uses the content of AI learning from the various real-time chat contents and response comments thereto as a dataset to generate response comments such as product information and purchase-inducing information that can induce the purchase of a product currently on sale, converts the generated response comments into voice, and enables the AI ​​show host to communicate with the viewer by playing them through the broadcast.

[0081] In this specification, only a few examples among the various embodiments carried out by the inventors are described; however, the technical concept of the present invention is not limited or restricted thereto, and it is obvious that it can be modified and implemented in various ways by those skilled in the art. Explanation of the symbols

[0082] 100: AI Show Host Generation Server 110: AI Show Host Generation Unit 120: Product Image Generation Unit 130: Broadcast Content Creation Department 200: Live Commerce Operation Server 210: Broadcast Content Correction Department 220: Live Commerce Execution Department 230: Viewer Specific Department 240: Viewer Communication Department

Claims

Claim 1 An AI show host generation server that generates an image of an AI show host who is the host of a live commerce broadcast and an image of a product to be sold, generates a broadcast scenario to be conducted by the AI ​​show host for product sales, and then generates broadcast content that sets the location information of the AI ​​show host moving on the broadcast screen and the location information of the product image according to the broadcast scenario, and stores it in a database; A live commerce operating server that specifies a country to transmit a live commerce broadcast, translates the broadcast scenario into the language of the country, corrects at least one of the mouth shape, facial expression, and gesture of the AI ​​show host according to the translated language, and transmits the corrected broadcast content over a communication network; wherein the AI ​​show host generation server, when generating the broadcast content, mutually matches the hand exposure location information of the AI ​​show host with the location information of the product image to display the product image at the position where the AI ​​show host is holding the product, or designates the area between the location information of the two hands of the AI ​​show host as a product display area, and generates the broadcast content to rotate the product image 360 ​​degrees within the product display area, and when the product is displayed in the area between the two hands, displays the product image separately or displays the internal composition of the product; and the live commerce operating server includes a viewer identification unit that identifies the viewer using the ID of the viewer who entered real-time chat during the live commerce, and queries the past live commerce participation history and product purchase history of the identified viewer from a database.A live commerce provision system using an AI show host, comprising: a viewer communication unit that generates corresponding text using a Large Language Model (LM) trained on response comment data to induce chatting and product purchase in past live commerce based on real-time chat input by the viewer and the viewed viewer's history; converts the generated corresponding text into voice data; and generates at least one of the mouth shape, facial expression, and gesture of the AI ​​show host synchronized with the voice data, thereby generating and transmitting feedback content in which the display state of the AI ​​show host image and the product image is changed. Claim 2 In claim 1, the AI ​​show host generation server comprises: an AI show host generation unit that generates an avatar having a human-like appearance, wherein the avatar's face is generated to have multiple facial expressions capable of expressing various emotions, and the avatar's hands are generated to have multiple hand movements capable of taking various gestures, including the action of grasping a product, and stores them in a database; a product image generation unit that generates products to be sold through live commerce by 3D scanning or stores various product images received from a manufacturer in a database; and a broadcast content generation unit that generates a broadcast scenario in which the AI ​​show host and the product image are displayed together on a broadcast screen to provide explanations and promotions of the product to viewers, specifies the facial expressions and gestures taken by the AI ​​show host displayed on the screen during live commerce according to the broadcast scenario, and sets location information of the product image displayed together with the AI ​​show host to generate broadcast content; a live commerce providing system using an AI show host. Claim 3 In paragraph 2, the AI ​​show host generation unit generates an AI show host by creating an avatar of a prototype show host having a basic human appearance, and then generating an AI show host by querying an appearance preference database pre-established for each country and correcting the show host's clothing, skin color, and face shape according to the country or region where the live commerce is to be broadcast. This is a live commerce provision system using an AI show host. Claim 4 delete Claim 5 delete Claim 6 In paragraph 2, the live commerce operating server comprises: a broadcast content correction unit that specifies a country to which a live commerce broadcast is to be transmitted, translates the broadcast scenario into the language of the country, and corrects the clothing, mouth shape, facial expression, voice tone, or gesture of the AI ​​show host according to the translated language to generate broadcast content for the country; and a live commerce execution unit that transmits the broadcast content for the country over a communication network so that the AI ​​show host promotes products in the language of each country and induces purchases; a live commerce providing system using an AI show host. Claim 7 In claim 6, the viewer communication unit conducts real-time communication based on LLM using comments entered by viewers, generates product information or purchase-related information desired by viewers or response comments that can induce purchase, and generates and plays back the voice, lip movements, or gestures of the AI ​​show host to output the generated response comments, thereby executing real-time two-way communication with viewers, a live commerce providing system using an AI show host. Claim 8 In claim 7, the viewer communication unit performs two-way communication by using the AI ​​show host, wherein the viewer communication unit uses a dataset of various types of chats entered by viewers in past live commerce and response comments that induce viewers to purchase products in response to them, and learns AI based on LLM (Large Language Model), and uses the learning results to generate and output response comments for comments registered during the current live commerce as feedback. Claim 9 In claim 8, the viewer communication unit corrects the facial expression or voice tone of the AI ​​show host according to the gender of the viewer, generates different response comments when the viewer is female and when the viewer is male, and performs individual responses to the viewer, a live commerce providing system using an AI show host. Claim 10 In claim 8, the viewer communication unit generates response comments for viewers currently watching the live commerce based on the post-purchase usage experiences of buyers of the same product stored in a database during a previous live commerce, thereby enabling empathy in sharing the product's usage experience, a live commerce provision system using an AI show host.

Images