Context-Aware Chatbot Extension Leveraging HTML Data and Retrieval-Augmented Generation (RAG)

Yang Ouyang, Tong Yu, Wenchu Wang

Keywords: Context - Aware, Chatbot Extension, HTML Data, Retrieval - Augmented Generation

TL;DR: Browser - based chatbot extension using HTML data and RAG, powered by a fine - tuned LLaMA 3 model, addresses context - awareness in chatbots for better web information retrieval.

Abstract:

In the digital age, the need for efficient information retrieval has become increasingly crucial. With
the vast amount of information available on the web, users often find themselves in situations where
they are seeking quick answers or additional information related to the current page they are viewing.
However, the process of switching between multiple pages or looking up external resources can be
disruptive and time- consuming, which significantly affects the user experience.Traditional chatbots,
while being useful to some extent, often lack the ability to adapt to the context. Specifically, they
struggle to offer responses directly based on a user’s past and present interactions with a website.
This shortcoming has led to a demand for more intelligent information retrieval systems.
To address these issues, the objective of this project is to develop a browser- based chatbot extension.
This chatbot will be designed to retrieve and synthesize information from two main sources: previously
parsed HTML data and traditional Retrieval- Augmented Generation (RAG) sources. By leveraging
both of these sources, the chatbot aims to provide accurate, context- sensitive responses. Moreover,
it will be powered by a fine- tuned LLaMA 3 model. This model, which is capable of handling
HTML context, is expected to enhance the user experience by providing contextual insights and
enabling more efficient information retrieval. In essence, this project bridges the gap between the
user’s need for quick and relevant information and the limitations of existing chatbot technologies by
integrating stored HTML context and external data sources, thus offering users streamlined access
to relevant information through a browser extension. This development is not only important for
improving the user experience in day- to- day web browsing but also has the potential to impact
the field of information retrieval and natural language processing, setting new standards for context
aware chatbot systems.

Submission Number: 31