Smoothing Entailment Graphs with Language Models
Preprint
Abstract
The diversity and Zipfian frequency distribution of natural language predicates in corpora leads to sparsity in Entailment Graphs (EGs) built by Open Relation Extraction (ORE). EGs are computationally efficient and explainable models of natural language inference, but as symbolic models, they fail if a novel premise or hypothesis vertex is missing at test-time. We present theory and methodology for overcoming such sparsity in symbolic models. First, we introduce a theory of optimal smoothing of EGs by constructing transitive chains. We then demonstrate an efficient, open-domain, and unsupervised smoothing method using an off-the-shelf Language Model to find approximations of missing premise predicates. This improves recall by 25.1 and 16.3 percentage points on two difficult directional entailment datasets, while raising average precision and maintaining model explainability. Further, in a QA task we show that EG smoothing is most useful for answering questions with lesser supporting text, where missing premise predicates are more costly. Finally, controlled experiments with WordNet confirm our theory and show that hypothesis smoothing is difficult, but possible in principle.Keywords
All Related Versions
This publication has 27 references indexed in Scilit:
- Modality and Negation in Event Extraction2021
- Incorporating Temporal Information in Entailment Graph Mining2020
- Do Neural Language Models Overcome Reporting Bias?2020
- On the Sentence Embeddings from Pre-trained Language Models2020
- Learning Typed Entailment Graphs with Global Soft ConstraintsTransactions of the Association for Computational Linguistics, 2018
- Do neural nets learn statistical laws behind natural language?PLOS ONE, 2017
- Efficient Global Learning of Entailment GraphsComputational Linguistics, 2015
- Combined Distributional and Logical SemanticsTransactions of the Association for Computational Linguistics, 2013
- Basic objects in natural categoriesCognitive Psychology, 1976
- Family resemblances: Studies in the internal structure of categoriesCognitive Psychology, 1975