The rapid expansion of digital communication and document management necessitates the development of advanced solutions that seamlessly integrate conversational interfaces with document processing capabilities. This paper introduces an Interactive Chat Application that leverages Large Language Model (LLM) technology to enrich conversations with PDF and text file content. The application provides functionalities such as question answering, streamlining information retrieval and enhancing collaborative discussions within a unified platform. Significant technical challenges, including balancing computational demands with low-latency communication, are addressed through effective management of LLM resources and performance optimization. The application holds potential across various sectors, including enterprise search, customer support, legal research, and education, promising to revolutionise internal document search processes and enhance AI-powered customer support. Additionally, it promotes social benefits such as improved accessibility, multilingual conversation facilitation, and the democratisation of knowledge. The evaluation of the application’s question-answering capabilities is conducted using ROUGE metrics and assessments of Human-Level Performance (HLP), demonstrating its effectiveness and accuracy in understanding and generating responses.

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

Interactive Conversational Chat Application for Question and Answering Enabled by a Large Language Model

  • Qi Zheng Tai,
  • Kai Le Kong,
  • Xue Wen Tan,
  • Chi Wee Tan,
  • Khai Yin Lim,
  • Gloria Jennis Tan

摘要

The rapid expansion of digital communication and document management necessitates the development of advanced solutions that seamlessly integrate conversational interfaces with document processing capabilities. This paper introduces an Interactive Chat Application that leverages Large Language Model (LLM) technology to enrich conversations with PDF and text file content. The application provides functionalities such as question answering, streamlining information retrieval and enhancing collaborative discussions within a unified platform. Significant technical challenges, including balancing computational demands with low-latency communication, are addressed through effective management of LLM resources and performance optimization. The application holds potential across various sectors, including enterprise search, customer support, legal research, and education, promising to revolutionise internal document search processes and enhance AI-powered customer support. Additionally, it promotes social benefits such as improved accessibility, multilingual conversation facilitation, and the democratisation of knowledge. The evaluation of the application’s question-answering capabilities is conducted using ROUGE metrics and assessments of Human-Level Performance (HLP), demonstrating its effectiveness and accuracy in understanding and generating responses.