Year: 2024
Type: dissertation
Abstract: The efficiency of accessing and managing project documentation is crucial in the evolving landscape of software development and project management. Traditional Document Management Systems (DMSs) face challenges in providing efficient and contextually relevant search results, often falling short in interpreting the nuances behind user queries and the contextual relevance of documents. To address these challenges, this research explores the integration of vector databases and Large Language Models (LLMs) to enable semantic search capabilities and enhance the user experience in DMSs. Vector databases store textual documentation as high-dimensional vectors, and when combined with the contextual comprehension provided by LLMs, the search functionality offers a more nuanced and effective method for query interpretation. This integration aims to ensure that the search results are accurate and contextually relevant. The scope of this thesis involves integrating these advanced technologies into a traditional DMS for project documentation. The experimental platform used is AggroDoc, which serves as an example to detail the necessary architectural modifcations and user interface improvements for such an integration. A qualitative case study is conducted to evaluate the usefulness of the enhanced functionality. The study focuses on how effectively participants can navigate and understand a topic with the help of semantic search capabilities. Results show how integrating advanced state-of-the-art technology for efficient and comprehensive document retrieval can be implemented and contribute to a better user experience, increasing productivity and engagement.
Author: Masoudi, Arian
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7017336767