Fact-driven Logical Reasoning

Abstract

Logical reasoning, which is closely related to human cognition, is of vital importance in human's understanding of texts. Recent years have witnessed increasing attentions on machine's logical reasoning abilities. However, previous studies commonly apply ad-hoc methods to model pre-defined relation patterns, such as linking named entities, which only considers global knowledge components that are related to commonsense, without local perception of complete facts or events. Such methodology is obviously insufficient to deal with complicated logical structures. Therefore, we argue that the natural logic units would be the group of 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, and enhance the supergraph with further explicit guidance of local question and option interactions. Experiments on two challenging logical reasoning benchmark datasets, ReClor and LogiQA, show that our proposed model, FOCAL REASONER, outperforms the baseline models dramatically. It can also be smoothly applied to other downstream tasks such as MuTual, a dialogue reasoning dataset, achieving competitive results.

Logiformer: A Two-Branch Graph Transformer Network for Interpretable Logical Reasoning

“…But it simply forms a chaintype discourse network and weakens the relations between two distant units. FocalReasoner [21] stresses that fact units in the form of subject-verb-object are significant for logical reasoning. It constructs a supergraph on top of the fact units and updates the node features relying on Graph Neural Network.…”

“…• DAGN [12]: It proposed a discourse-aware network, which took RoBERTa-Large [19] as the token encoder and employed GNN for the feature update. • FocalReasoner [21]: It focused on the fact units extracted from the text and built a supergraph for the reasoning. Similar to DAGN, it also leveraged RoBERTa-Large and GNN [26] for the token embedding and node update respectively.…”

“…However, it still embeds the logic in an implicit form and fails to make up for the weakness of PLMs in logical reasoning. Some works like FocalReasoner [21] uncovers the logical structure in only one aspect (e.g., capture the co-occurrence between units). It leads to the weak capability of the model to capture logical relationships.…”

LogiTorch: A PyTorch-based library for logical reasoning on natural language

“…He won the World Cup in 1998", the words "Zidane" and "he" refer to the same person. Coreference resolution is used by the Focal Reasoner model (Ouyang et al, 2021) to construct a graph of fact triples, where the same mentions are connected with an undirected edge. In LogiTorch, we implemented a wrapper over a finetuned SpanBERT (Joshi et al, 2020) for coreference resolution.…”

“…This model is more robust on examples containing negations, and performs better on the negated NLI dataset than the original BERT. Future releases will include newer models such as LReasoner (Huang et al, 2021), Focal Reasoner (Ouyang et al, 2021), AdaLoGN (Li et al, 2022), Logiformer (Xu et al, 2022), and LogiGAN (Pi et al, 2022).…”