Answer Consolidation: Formulation and Benchmarking
This addresses a real-world need in QA applications for handling multiple answer scenarios, but it is incremental as it formulates and benchmarks a new task rather than introducing a breakthrough method.
The paper tackles the problem of answer consolidation in question answering, where multiple answers need to be grouped into aspects and consolidated into a non-redundant set, and they constructed a dataset of 4,699 questions and 24,006 sentences to benchmark models, with the best-performing supervised models showing promising but improvable results.
Current question answering (QA) systems primarily consider the single-answer scenario, where each question is assumed to be paired with one correct answer. However, in many real-world QA applications, multiple answer scenarios arise where consolidating answers into a comprehensive and non-redundant set of answers is a more efficient user interface. In this paper, we formulate the problem of answer consolidation, where answers are partitioned into multiple groups, each representing different aspects of the answer set. Then, given this partitioning, a comprehensive and non-redundant set of answers can be constructed by picking one answer from each group. To initiate research on answer consolidation, we construct a dataset consisting of 4,699 questions and 24,006 sentences and evaluate multiple models. Despite a promising performance achieved by the best-performing supervised models, we still believe this task has room for further improvements.