Fact-driven Logical Reasoning

Siru Ouyang, Zhuosheng Zhang, hai zhao

Keywords: logical reasoning, machine reading comprehension, language understanding

Abstract: Recent years have witnessed an increasing interest in training machines with reasoning ability, which deeply relies on accurate, clearly presented clue forms that are usually modeled as entity-like knowledge in existing studies. However, in real hierarchical reasoning motivated machine reading comprehension, such one-sided modeling is insufficient for those indispensable local complete facts or events when only "global" knowledge is really paid attention to. Thus, in view of language being a complete knowledge/clue carrier, we propose a general formalism to support representing logic units by extracting backbone constituents of the sentence such as the subject-verb-object formed "facts", covering both global and local knowledge pieces that are necessary as the basis for logical reasoning. Beyond building the ad-hoc graphs, we propose a more general and convenient fact-driven approach to construct a supergraph on top of our newly defined fact units, benefiting from both sides of the connections between facts and internal knowledge such as concepts or actions inside a fact. Experiments on two challenging logical reasoning benchmarks show that our proposed model, \textsc{Focal Reasoner}, outperforms the baseline models dramatically and achieves state-of-the-art results.

Community Implementations: [ 2 code implementations](https://www.catalyzex.com/paper/arxiv:2105.10334/code)