Search

CN-121998069-A - Marketing display method and system based on multiple agents

CN121998069ACN 121998069 ACN121998069 ACN 121998069ACN-121998069-A

Abstract

The application provides a marketing display method and a marketing display system based on multiple agents. The method and the system can provide customized display schemes for different client types, improve the transmission efficiency of effective information in the marketing display process, and improve the user experience. In addition, the embodiment of the application can also improve the quality of the products, has the advantages of strong interpretability and expandability, and can also support demonstration, question-answering and the like in the professional field.

Inventors

  • MA YUANCHEN
  • LI LAN
  • ZHENG MENG

Assignees

  • 株式会社日立制作所

Dates

Publication Date
20260508
Application Date
20241108

Claims (14)

  1. 1. A multi-agent based marketing display method, comprising: a customer survey agent generates a customer survey report including customer types, customer portraits for each customer type, or customer characteristics; the product investigation agent generates a product investigation report of the target product; a material editing agent generates a product explanation material matched with each customer type based on the customer survey report and the product survey report; the digital demonstration agent selects the product explanation materials matched with the on-site client to demonstrate according to the client type of the on-site client.
  2. 2. The method of claim 1, wherein the customer survey agent generates a customer survey report comprising: Acquiring client-related information, wherein the client-related information comprises at least one of exhibition names, determined participant lists, predicted participant lists, determined participant lists and predicted participant lists; Based on the client related information, obtaining client attributes, wherein the client attributes comprise at least one of age, gender, used language type, affiliated units, occupation, position, technical personnel or business personnel; based on the client attributes, a client type of the client and a client representation or client characteristic for each client type is determined.
  3. 3. The method of claim 1, wherein the product survey agent generating a product survey report for the target product comprises: And based on the existing product knowledge base, the target product is investigated, and a product investigation report of the target product is generated, wherein the product investigation report comprises at least one product information item, and the product information item comprises background knowledge, functions, application scenes, application schemes, commercial value, dominant technology and future plans.
  4. 4. The method of claim 1, wherein the material editing agent generates product narrative material matching each customer type based on the customer survey report and the product survey report, including at least one of a first collaboration mode, a second collaboration mode, a third collaboration mode, and a fourth collaboration mode, wherein, In the first collaboration mode, the customer survey agent and the product survey agent each independently generate a survey report; the material editing agent makes product explanation materials matched with each customer type based on investigation reports generated by the customer investigation agent and the product investigation agent independently; In the second collaboration mode, the material editing intelligent body and the product investigation intelligent body respectively generate investigation reports and/or answers according to investigation requests and/or questions sent by the material editing intelligent body in at least one interaction process; in the third collaboration mode, the customer investigation agent and the product investigation agent respectively generate corresponding investigation reports according to task scheduling of project manager agents, and the material editing agent generates product explanation materials matched with each customer type based on the investigation reports according to the task scheduling of the project manager agents; The fourth collaboration mode is that in the first collaboration mode, the second collaboration mode or the third collaboration mode, after a first intermediate result is generated, a relevant agent receives correction and/or confirmation of the first intermediate result by a user, and executes a next task according to the first intermediate result after the user correction or confirmation, wherein the first intermediate result comprises at least one of a product investigation report, a customer investigation report, a problem generated intelligently by editing materials, an answer generated by the material editing agent or the product investigation agent, and a outline or chapter of a comment material.
  5. 5. The method as recited in claim 1, further comprising: The materials editing agent further generates a commentary suggestion matching each customer type when generating a product commentary material matching each customer type based on the customer survey report and the product survey report.
  6. 6. The method as recited in claim 5, further comprising: aiming at each client type, the virtual client agent and the virtual explanation agent generate or update a question-answer knowledge base corresponding to the client type through at least one round of question-answer interaction process; In the question-answer interaction process, the virtual client agent simulates a client of the client type according to the client portrait or the client characteristic of the client type and presents a question based on product explanation materials matched with the client type, and generates an answer to the question presented by the virtual client agent based on at least one of a product survey report, the product explanation materials, the explanation suggestions and a product knowledge base.
  7. 7. The method as recited in claim 6, further comprising: after each round of the question-answer interaction process, for each customer type, the material editing agent updates product commentary material and/or commentary advice matching the customer type based on the question-answer knowledge base corresponding to the customer type.
  8. 8. The method of claim 7, wherein the product commentary material and/or commentary suggested update modes, including at least one of a first update mode, a second update mode, a third update mode and a fourth update mode, In the first updating mode, the virtual client agent and the virtual explanation agent are the same agent, and a question-answer knowledge base corresponding to the client type is generated through a ask oneself self-answer mode; The material editing agent updates the product explanation materials and/or the explanation suggestions based on the information obtained by interaction of the question-answer knowledge base, the client investigation agent and the product investigation agent; In the third updating mode, the virtual client agent and the virtual explanation agent are different agents; the virtual client agent and the virtual explanation agent execute a question-answer interaction process according to task scheduling of the project manager agent and generate the question-answer knowledge base; the material editing agent updates the product explanation materials and/or explanation suggestions based on the question-answer knowledge base according to the task schedule of the project manager agent; The fourth updating mode is that after the second intermediate result is generated in the first updating mode, the second updating mode or the third updating mode, the relevant agent receives the correction and/or confirmation of the second intermediate result by the user and executes the next task according to the second intermediate result corrected or confirmed by the user, wherein the second intermediate result comprises at least one of a question generated by the virtual client agent, an answer generated by the virtual explanation agent, a modification mode of the product explanation material and/or the explanation suggestion.
  9. 9. The method of claim 1, wherein, Selecting a digital person image and a demonstration mode matched with the client type of the on-site client, and demonstrating the product explanation materials matched with the on-site client according to the selected demonstration mode by the selected digital person image, wherein the demonstration mode comprises on-site demonstration and/or playing of the product explanation materials; under the condition that the product explanation materials are not matched with the target language used by the on-site clients, the explanation words of the digital demonstration intelligent body are converted into the target language and displayed; When the digital demonstration agent selects the product explanation materials matched with the on-site client to demonstrate according to the client type of the on-site client, the digital demonstration agent further selects the explanation suggestions matched with the on-site client, and demonstrates the product explanation materials according to the selected explanation suggestions.
  10. 10. The method as recited in claim 1, further comprising: The material editing agent generates a product narrative material matching a fusion type based on the customer survey report and the product survey report, the fusion type including at least two customer types, the at least two customer types being predetermined or determined from a customer type of an on-site customer.
  11. 11. The method of claim 6 or 7, further comprising: the feedback collection agent performs question-answer interaction with the on-site client based on the question-answer knowledge base and collects questions, answers and feedback comments of the on-site client in the interaction process, wherein when the feedback collection agent cannot answer or answer cannot meet the requirements of the on-site client, the feedback collection agent collects answers provided by on-site real person explanation personnel; The feedback collection agent updates the question-answer knowledge base according to the collected questions, answers and feedback comments of the site clients and provides the question-answer knowledge base to the material editing agent and/or the digital demonstration agent; the material editing agent updates the product commentary material and/or commentary suggestion based on the updated question-answer knowledge base.
  12. 12. The method of claim 6 or 7, further comprising: The live recording agent records interaction information in the demonstration process and generates a meeting summary and/or a customer feedback report corresponding to the customer type.
  13. 13. Marketing display system based on many agents, its characterized in that includes customer investigation agent, product investigation agent, material editing agent and digital demonstration agent, wherein: The customer survey agent is used for generating a customer survey report, wherein the customer survey report comprises customer types, customer portraits of the customer types or customer characteristics; the product investigation intelligent agent is used for generating a product investigation report of a target product; The material editing agent is used for generating product explanation materials matched with each customer type based on the customer survey report and the product survey report; The digital demonstration intelligent agent is used for selecting product explanation materials matched with the on-site clients to demonstrate according to the client types of the on-site clients.
  14. 14. A computer program product comprising computer instructions which, when executed by a processor, implement the steps of the method of any of claims 1 to 12.

Description

Marketing display method and system based on multiple agents Technical Field The application relates to the technical field of artificial intelligence (ARTIFICIAL INTELLIGENCE, AI), in particular to a marketing display method and system based on multiple agents. Background Question-answering systems based on Large Language Models (LLM) are widely used in natural language processing, intelligent customer service, intelligent home and many other fields. The above-described system may be deficient in some specific application scenarios. Taking the introduction of products or proposals for potential customers at an exhibition or business meeting as an example, for various reasons (such as lack of enough knowledge of the customer's background in advance), the visiting customers may not be able to be specifically introduced in the field, or it may be difficult to make adequate preparation for possible questions of the customers, resulting in low efficiency of effective information transmission, difficulty in meeting the user's needs, and limited promotion effect on business cooperation. One prior art (refer to chinese patent application No. 202310979105.5) parses out the model identity required to answer the question from the user's request, then retrieves the corresponding Agent information (e.g., identity description information such as history specialists, doctors, etc.) from the database, generates a prompt term (prompt) in combination with the user's original question, and then generates a corresponding answer in a virtual specific identity by LLM. The prior art can answer questions of users in a targeted way, improve the experience of users, but has limitations, and at least comprises the following aspects: 1. The answers are customized based only on a single question posed by the user, without regard to the actual context of the different users. Users of different contexts will get the same feedback when asking the same question. 2. No targeted adaptation and customization is performed based on specific knowledge bases such as products or proposals to be demonstrated. 3. Only suitable for real-time communication, and completely depends on LLM and is not participated by manpower. Considering the limitations of current AI technology, especially in the field of lecturing professionals, AI often cannot completely replace a real person lecturer. The above-described technique does not facilitate adequate preparation by the presenter in advance. 4. The description is not considered to be iterated or summarized during or after the process. Disclosure of Invention At least one embodiment of the application provides a marketing display method and a marketing display system based on multiple agents, which can provide customized display schemes for different client types and improve the transmission efficiency of effective information in the marketing display process. According to a first aspect of the present application, at least one embodiment provides a multi-agent-based marketing display method, including: a customer survey agent generates a customer survey report including customer types, customer portraits for each customer type, or customer characteristics; the product investigation agent generates a product investigation report of the target product; a material editing agent generates a product explanation material matched with each customer type based on the customer survey report and the product survey report; the digital demonstration agent selects the product explanation materials matched with the on-site client to demonstrate according to the client type of the on-site client. Optionally, the client survey agent generates a client survey report, including: Acquiring client-related information, wherein the client-related information comprises at least one of exhibition names, determined participant lists, predicted participant lists, determined participant lists and predicted participant lists; Based on the client related information, obtaining client attributes, wherein the client attributes comprise at least one of age, gender, used language type, affiliated units, occupation, position, technical personnel or business personnel; based on the client attributes, a client type of the client and a client representation or client characteristic for each client type is determined. Optionally, the product survey agent generates a product survey report of the target product, including: And based on the existing product knowledge base, the target product is investigated, and a product investigation report of the target product is generated, wherein the product investigation report comprises at least one product information item, and the product information item comprises background knowledge, functions, application scenes, application schemes, commercial value, dominant technology and future plans. Optionally, the material editing agent generates a product description material matching each customer type based o