Representations for Question Answering from Documents with Tables and Text

Tables in Web documents are pervasive and can be directly used to answer many\nof the queries searched on the Web, motivating their integration in question\nanswering. Very often information presented in tables is succinct and hard to\ninterpret with standard language representations. On the other hand, tables\noften appear within textual context, such as an article describing the table.\nUsing the information from an article as additional context can potentially\nenrich table representations. In this work we aim to improve question answering\nfrom tables by refining table representations based on information from\nsurrounding text. We also present an effective method to combine text and\ntable-based predictions for question answering from full documents, obtaining\nsignificant improvements on the Natural Questions dataset.\n