Multi-Step Inference for Reasoning Over Paragraphs

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 3040–3050, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 3040 Multi-Step Inference for Reasoning Over Paragraphs Jiangming Liu♠∗, Matt Gardner♣, Shay B. Cohen♠, Mirella Lapata♠ ♠ ILCC, University of Edinburgh ♣ Allen Institute for AI jiangming.liu@ed.ac.uk, mattg@allenai.org, {scohen,mlap}@inf.ed.ac.uk Abstract Complex reasoning over text requires under- standing and chaining together free-form pred- icates and logical connectives. Prior work has largely tried to do this either symbolically or with black-box transformers. We present a middle ground between these two extremes: a compositional model reminiscent of neural module networks that can perform chained log- ical reasoning. This model first finds relevant sentences in the context and then chains them together using neural modules. Our model gives significant performance improvements (up to 29% relative error reduction when com- bined with a reranker) on ROPES, a recently- introduced complex reasoning dataset. 1 Introduction Performing chained inference over natural lan- guage text is a long-standing goal in artificial in- telligence (Grosz et al., 1986; Reddy, 2003). This kind of inference requires understanding how nat- ural language statements fit together in a way that permits drawing conclusions. This is very chal- lenging without a formal model of the semantics underlying the text, and when polarity needs to be tracked across many statements. For instance, consider the example in Figure 1 from ROPES (Lin et al., 2019), a recently released reading comprehension dataset that requires apply- ing information contained in a background para- graph to a new situation. To answer the question, one must associate each category of flowers with a polarity for having brightly colored petals, which must be done by going through the information about pollinators given in the situation and linking it to what was said about pollinators and brightly colored petals in the background paragraph, along with tracking the polarity of those statements. ∗ Work done during an internship at Allen Institute for Artificial Intelligence. Background: Scientists think that the earliest flowers attracted insects and other animals, which spread pollen from flower to flower. This greatly increased the efficiency of fertilization over wind-spread pollen, which might or might not actually land on another flower. To take bet- ter advantage of this animal labor , plants evolved traits such as brightly colored petals to attract pollinators. In exchange for pollination, flowers gave the pollinators nec- tar. Situation: Last week, John visited the national park near his city. He saw many flowers. His guide explained him that there are two categories of flowers, category A and category B. Category A flowers spread pollen via wind, and category B flowers spread pollen via animals. Question: Which category of flowers would be more likely to have brightly colored petals? Answer: category B (a) SELECT CHAIN CHAIN PREDICT background question situation category B (b) Figure 1: (a) An example in ROPES; (b) the chained reasoning that our model performs on the example. The model first (softly) selects relevant parts of the back- ground and question, then …

Content truncated to 500 words