SemExplorer: A User Interface for Semantic Approach to Customized Dataset Search
Zixin Wei, Jun Gang Han, Xiaolin Han, Chenhao Ma · 2025
In response to the increasing complexity of data and the widespread availability of diverse datasets for modeling complex relationships, there is a crucial need for efficient and customizable search capabilities. Many data scientists and researchers still need to manually browse through extensive catalogs to find suitable datasets for their studies. To address this challenge, we introduce SemExplorer, a framework designed to enhance semantic analysis and information retrieval for user-defined needs. Our system processes and interprets complex queries in natural language using large language models (LLMs), converting them into vector representations and statistical filters. When presenting search results, our system uses context dependencies to highlight important information, helping users quickly locate the results they need. Overall, SemExplorer is a robust tool that not only improves the efficiency of finding needed datasets but also enhances the precision of search outcomes by leveraging semantic analysis to deeply understand and filter network data. A demo of our system is available online.