TSQA: Tabular Scenario Based Question Answering

Scenario-based question answering (SQA) has attracted an increasing research interest. Compared with the well-studied machine reading comprehension (MRC), SQA is a more challenging task: scenario may contain not only textual passage to read but also structured data like tables, i.e., tabular based (TSQA). AI applications of TSQA such as multiple-choice questions in high-school exams require synthesizing multiple cells and combining tables texts domain knowledge infer answers. To support study this task, we construct GeoTSQA. This dataset contains 1k real contextualized by scenarios geography domain. solve extend state-of-the-art MRC methods TTGen, novel table-to-text generator. It generates sentences from variously synthesized feeds downstream method most useful sentences. Its sentence ranking model fuses information scenario, question, knowledge. Our approach outperforms variety strong baseline on