Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Learnable Agent Collaboration Network Framework for Personalized Multimodal AI Search Engine

Yunxiao Shi, Min Xu, Haimin Zhang, Xing Zi, Qiang Wu

TL;DR

A novel AI Search Engine framework called the Agent Collaboration Network (ACN), which consists of multiple specialized agents working collaboratively, each with distinct roles such as Account Manager, Solution Strategist, Information Manager, and Content Creator, enhancing the AI search engine's response quality, personalization, and interactivity.

Abstract

Large language models (LLMs) and retrieval-augmented generation (RAG) techniques have revolutionized traditional information access, enabling AI agent to search and summarize information on behalf of users during dynamic dialogues. Despite their potential, current AI search engines exhibit considerable room for improvement in several critical areas. These areas include the support for multimodal information, the delivery of personalized responses, the capability to logically answer complex questions, and the facilitation of more flexible interactions. This paper proposes a novel AI Search Engine framework called the Agent Collaboration Network (ACN). The ACN framework consists of multiple specialized agents working collaboratively, each with distinct roles such as Account Manager, Solution Strategist, Information Manager, and Content Creator. This framework integrates mechanisms for picture content understanding, user profile tracking, and online evolution, enhancing the AI search engine's response quality, personalization, and interactivity. A highlight of the ACN is the introduction of a Reflective Forward Optimization method (RFO), which supports the online synergistic adjustment among agents. This feature endows the ACN with online learning capabilities, ensuring that the system has strong interactive flexibility and can promptly adapt to user feedback. This learning method may also serve as an optimization approach for agent-based systems, potentially influencing other domains of agent applications.

A Learnable Agent Collaboration Network Framework for Personalized Multimodal AI Search Engine

TL;DR

A novel AI Search Engine framework called the Agent Collaboration Network (ACN), which consists of multiple specialized agents working collaboratively, each with distinct roles such as Account Manager, Solution Strategist, Information Manager, and Content Creator, enhancing the AI search engine's response quality, personalization, and interactivity.

Abstract

Large language models (LLMs) and retrieval-augmented generation (RAG) techniques have revolutionized traditional information access, enabling AI agent to search and summarize information on behalf of users during dynamic dialogues. Despite their potential, current AI search engines exhibit considerable room for improvement in several critical areas. These areas include the support for multimodal information, the delivery of personalized responses, the capability to logically answer complex questions, and the facilitation of more flexible interactions. This paper proposes a novel AI Search Engine framework called the Agent Collaboration Network (ACN). The ACN framework consists of multiple specialized agents working collaboratively, each with distinct roles such as Account Manager, Solution Strategist, Information Manager, and Content Creator. This framework integrates mechanisms for picture content understanding, user profile tracking, and online evolution, enhancing the AI search engine's response quality, personalization, and interactivity. A highlight of the ACN is the introduction of a Reflective Forward Optimization method (RFO), which supports the online synergistic adjustment among agents. This feature endows the ACN with online learning capabilities, ensuring that the system has strong interactive flexibility and can promptly adapt to user feedback. This learning method may also serve as an optimization approach for agent-based systems, potentially influencing other domains of agent applications.
Paper Structure (27 sections, 9 figures, 1 algorithm)

This paper contains 27 sections, 9 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. AI Search Engine
  4. Personalized Generation of LLM
  5. AI Agent
  6. Agent Collaboration Network
  7. Account Manager
  8. Solution Strategist
  9. Information Manager
  10. Content Creator
  11. Role Setting
  12. Prompt for Account Manager
  13. Prompt for Solution Strategist
  14. Prompt for Content Creator
  15. Reflective Forward Optimization
  16. ...and 12 more sections

Figures (9)

  • Figure 1: Agent Collaboration Network Framework with a case study "Place to buy cat using product." The red text illustrates how the ACN customizes message passing, the information searching process, and the information generation process to the specific user. The mind map is generated using COT, guiding the function calling for solving the user's information gaining requirements in a logic, deep, and structural way.
  • Figure 2: RFO algorithm workflow.
  • Figure 3: Comparison of AI Search Engine Responses to the Query "Give me a dietary recommendation for building muscle." A LLM played judge subsequently determines that Response B (ACN) is better.
  • Figure 4: The results of pairwise comparisons between Basic and ACN responses across all categories on the MSMTPInfo dataset.
  • Figure 5: Comparison between ACN and Perplexity.
  • ...and 4 more figures