Open Domain Question Answering over Tables via Dense Retrieval
This addresses the problem of retrieving and answering questions from tables for users in information-seeking domains, representing an incremental advance by adapting dense retrieval to a new data type.
The paper tackles open-domain question answering over tables by introducing a dense retriever designed for tabular context, improving retrieval recall@10 from 72.0 to 81.1 and exact match QA results from 33.8 to 37.7.
Recent advances in open-domain QA have led to strong models based on dense retrieval, but only focused on retrieving textual passages. In this work, we tackle open-domain QA over tables for the first time, and show that retrieval can be improved by a retriever designed to handle tabular context. We present an effective pre-training procedure for our retriever and improve retrieval quality with mined hard negatives. As relevant datasets are missing, we extract a subset of Natural Questions (Kwiatkowski et al., 2019) into a Table QA dataset. We find that our retriever improves retrieval results from 72.0 to 81.1 recall@10 and end-to-end QA results from 33.8 to 37.7 exact match, over a BERT based retriever.