Unified Question Answering
Project 2023-24
Project Title: Unified Question Answering
Lab/Research Group:
Center for Intelligent Information Retrieval (CIIR) is a world leading research lab in information retrieval. CIIR's projects include many aspects of developing and evaluating search engines, question answering systems, and recommender systems. The current projects of the Center include but are not limited to neural information retrieval, conversational information seeking, controversy detection, and cross-lingual information retrieval. For more information, visit the Center's website at http://ciir.cs.umass.edu/
Project Description:
Question answering (QA) is at the heart of information access systems. Different methods, datasets, and evaluation methodologies have been developed for various question answering systems, such as factoid QA, non-factoid QA, visual QA, etc. The goal of this project is to unify these efforts and train a single deep learning models for all these QA problems.
Learning Objectives:
Students should be able to train a model on at least three different QA setups and explore the performance of the unified model on unseen test datasets.
Skills needed:
Proficiency in Python and basic knowledge of machine learning are required for this project.