Scholarlink

This paper is devoted to the development and implementation of an artificial intelligence (AI)-based system for enhancing the interaction between human and computer through natural language, text-based document search, and gesture-based interface. To accomplish this, the proposed methodology employs artificial intelligence models and private data sources and employs Retrieval-Augmented Generation (RAG) models to make it a dynamic and flexible system. The chatbot developed with the help of Llama-2-7B-Chat model provides good responses to the user’s queries. On the other hand, the RAG model leveraged on Gemini Pro of Google AI to provide relevant and pertinent answers from uploaded documents in different formats. To make the usage of these components as smooth and intuitive, they are incorporated into a cross-platform desktop application. This paper proves the effectiveness and possibility of integrating several AI technologies into one tool. It also provides guidance on how to advance on the development of personalized artificial intelligence systems, which can be applicable in business, education, and productivity.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced Human-Computer Interaction with Artificial Intelligence Models Based on Retrieval-Augmented Generation

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ayman Aljarbouh,
  • Shahrom Siddiqzoda

摘要

This paper is devoted to the development and implementation of an artificial intelligence (AI)-based system for enhancing the interaction between human and computer through natural language, text-based document search, and gesture-based interface. To accomplish this, the proposed methodology employs artificial intelligence models and private data sources and employs Retrieval-Augmented Generation (RAG) models to make it a dynamic and flexible system. The chatbot developed with the help of Llama-2-7B-Chat model provides good responses to the user’s queries. On the other hand, the RAG model leveraged on Gemini Pro of Google AI to provide relevant and pertinent answers from uploaded documents in different formats. To make the usage of these components as smooth and intuitive, they are incorporated into a cross-platform desktop application. This paper proves the effectiveness and possibility of integrating several AI technologies into one tool. It also provides guidance on how to advance on the development of personalized artificial intelligence systems, which can be applicable in business, education, and productivity.