On the Summarization of Consumer Health Questions

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 2228–2234 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 2228 On the Summarization of Consumer Health Questions Asma Ben Abacha Dina Demner-Fushman asma.benabacha@nih.gov ddemner@mail.nih.gov LHNCBC, U.S. National Library of Medicine, Bethesda, MD Abstract Question understanding is one of the main challenges in question answering. In real world applications, users often submit natu- ral language questions that are longer than needed and include peripheral information that increases the complexity of the question, lead- ing to substantially more false positives in an- swer retrieval. In this paper, we study neural abstractive models for medical question sum- marization. We introduce the MeQSum corpus of 1,000 summarized consumer health ques- tions. We explore data augmentation meth- ods and evaluate state-of-the-art neural ab- stractive models on this new task. In particu- lar, we show that semantic augmentation from question datasets improves the overall perfor- mance, and that pointer-generator networks outperform sequence-to-sequence attentional models on this task, with a ROUGE-1 score of 44.16%. We also present a detailed error anal- ysis and discuss directions for improvement that are specific to question summarization. 1 Introduction Teaching machines how to automatically under- stand natural language questions to retrieve rele- vant answers is still a challenging task. Different factors increase the complexity of the task such as the question length (cf. Figure 1), the lexical heterogeneity when describing the same informa- tion need, and the lack of domain-specific training datasets. Improving Question Answering (QA) has been the focus of multiple research efforts in recent years. Several efforts proposed inter- active and non-interactive query relaxation tech- niques to translate the input questions into struc- tured queries covering specific elements of the questions (Yahya et al., 2013; Mottin et al., 2014; Ben Abacha and Zweigenbaum, 2015; Meng et al., 2017). Other efforts focused on (i) identifying question similarity (Nakov et al., 2016, 2017) and Figure 1: Consumer health questions and associated summaries from the gold standard. The entities in Red are the foci (main entities). The words in Blue and un- derlined are the triggers of the question types. question entailment (Ben Abacha and Demner- Fushman, 2019b) in order to retrieve similar or en- tailed questions that have associated answers, or (ii) paraphrasing the questions and submitting the simplified versions to QA systems (Bordes et al., 2014; Dong et al., 2017). Question simplification or summarization was less studied than the summarization of news ar- ticles that has been the focus of neural abstrac- tive methods in recent years (Rush et al., 2015; Nallapati et al., 2016; Chopra et al., 2016; See et al., 2017). In this paper, we tackle the task of consumer health question summarization. Con- sumer health questions are a natural candidate for this task as patients and their families tend to provide numerous peripheral details such as the patient history (Roberts and Demner-Fushman, 2016), that are not always needed to find cor- rect answers. Recent experiments also showed the 2229 …

Content truncated to 500 words