Arnold Bova, Paul Brouillette, Nicholas Morgan, Jaclyn Wirth

Abstract:

This research was done for TREC CAsT in which we worked on building a reranker system for the project. TREC is the Text Retrieval Conference sponsored by NIST. This specific track is CAsT, or Conversational Assistance Track. In this project, our goal was to return n items from a data set based on a query with the first result being the most relevant. For our part of the project, we worked on the reranker, in which passage chunks from documents would get reranked based on their relevance to the query.