Tag-based Multi-Span Extraction in Reading Comprehension

By Elad Segal, Avia Efrat, Mor Shoham, Amir Globerson, Jonathan Berant. 2019

30 citations recorded by the source corpus.

With models reaching human performance on many popular reading comprehension datasets in recent years, a new dataset, DROP, introduced questions that were expected to present a harder challenge for reading comprehension models. Among these new types of questions were "multi-span questions", questions whose answers consist of several spans from either the paragraph or the question itself. Until now, only one model attempted to tackle multi-span questions as a part of its design. In this work, we suggest a new approach for tackling multi-span questions, based on sequence tagging, which differs from previous approaches for answering span questions. We show that our approach leads to an absolute improvement of 29.7 EM and 15.1 F1 compared to existing state-of-the-art results, while not hurting performance on other question types. Furthermore, we show that our model slightly eclipses the current state-of-the-art results on the entire DROP dataset.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Neural Module Networks for Reasoning over Text (2019, methodology): “We compare to publicly available best performing models: NAQANet (Dua et al., 2019), NABERT+ (Kinley & Lin, 2019), TAG-NABERT+ (Avia Efrat & Shoham, 2019), and MTMSN (Hu et al., 2019), all trained on the same data as our model.”
Text Modular Networks: Learning to Decompose Tasks in the Language of Existing Models (2021, background): “To handle the former class, the first challenge is building models that can return a list of answers—a relatively unexplored task until recently (Hu et al., 2019b; Segal et al., 2020).”
Question Directed Graph Attention Network for Numerical Reasoning over Text (2020, background): “For multiple spans extraction, the probability is constructed referring to (Efrat et al., 2019).”
Towards Question Format Independent Numerical Reasoning: A Set of Prerequisite Tasks (2020, methodology): “Examples include NumNet+v2 [Ran et al. , 2019], BERT Calculator [Andor et al. , 2019] and tag-based multi-span extraction model [Efrat et al. , 2019].”
OPERA: Operation-Pivoted Discrete Reasoning over Text (2022, methodology): “Multi-spans Inspired by Segal et al. (2020), the multi-span answer (a set of non-contiguous spans) is derived with a sequence labeling method, in which each token of the input is tagged with BIO labels.”
What’s in Your Head? Emergent Behaviour in Multi-Task Transformer Models (2021, methodology): “Transformer models with multiple output heads have been widely employed in previous works (Hu 4Wilcoxon signed-rank test with p-value 0.001. and Singh, 2021; Aghajanyan et al., 2021; Segal et al., 2020; Hu et al., 2019; Clark et al., 2019).”