Neural Conversational QA: Learning to Reason v.s. Exploiting Patterns

Neural Conversational QA tasks like ShARC require systems to answer questions based on the contents of a given passage. On studying recent state-of-the-art models on the ShARCQA task, we found indications that the models learn spurious clues/patterns in the dataset. Furthermore, we show that a heuristic-based program designed to exploit these patterns can have performance comparable to that of the neural models. In this paper we share our findings about four types of patterns found in the ShARC corpus and describe how neural models exploit them. Motivated by the aforementioned findings, we create and share a modified dataset that has fewer spurious patterns, consequently allowing models to learn better.
Comment: Accepted at EMNLP 2020. NOTE: An older version of this paper presented a model called 'UrcaNet'. Please view the v1 version of this paper on arxiv for details on that model. This version does not contain UrcaNet