Interpretable Question Answering on Knowledge Bases and Text

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4943–4951 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 4943 Interpretable Question Answering on Knowledge Bases and Text Alona Sydorova iteratec GmbH Munich, Germany alona.sydorova@iteratec.com Nina Poerner & Benjamin Roth Center for Information and Language Processing LMU Munich, Germany {poerner,beroth}@cis.lmu.de Abstract Interpretability of machine learning (ML) models becomes more relevant with their in- creasing adoption. In this work, we address the interpretability of ML based question answer- ing (QA) models on a combination of knowl- edge bases (KB) and text documents. We adapt post hoc explanation methods such as LIME and input perturbation (IP) and com- pare them with the self-explanatory attention mechanism of the model. For this purpose, we propose an automatic evaluation paradigm for explanation methods in the context of QA. We also conduct a study with human annotators to evaluate whether explanations help them iden- tify better QA models. Our results suggest that IP provides better explanations than LIME or attention, according to both automatic and hu- man evaluation. We obtain the same ranking of methods in both experiments, which sup- ports the validity of our automatic evaluation paradigm. 1 Introduction Question answering (QA) is an important task in natural language processing and machine learn- ing with a wide range of applications. QA sys- tems typically use either structured information in the form of knowledge bases (KBs), or raw text. Recent systems have successfully combined both types of knowledge (Das et al., 2017). Nowadays, due to the changing legal situation and growing application in critical domains, ML based systems are increasingly required to pro- vide explanations of their output. Lipton (2018), Poursabzi-Sangdeh et al. (2018) and Doshi-Velez and Kim (2017) point out that there is no com- plete agreement on the definition, measurability and evaluation of interpretability in ML models. Nevertheless, a number of explanation methods have been proposed in the recent literature, with the aim of making ML models more transparent for humans. To the best of our knowledge, the problem of explanations for deep learning based QA models working on a combination of structured and un- structured data has not yet been researched. Also, there is a lack of evaluation paradigms to com- pare different explanation methods in the context of QA. Contributions - We explore interpretability in the context of QA on a combination of KB and text. In par- ticular, we apply attention, LIME and input perturbation (IP). - In order to compare these methods, we pro- pose a novel automatic evaluation scheme based on “fake facts”. - We evaluate whether explanations help hu- mans identify the better out of two QA mod- els. - We show that the results of automatic and hu- man evaluation agree. - Our results suggest that IP performs better than attention and LIME in this context. 2 Question Answering on Knowledge Bases and Text The combination of knowledge bases and text data is of particular interest in the context …

Content truncated to 500 words