Bilinear Fusion of Commonsense Knowledge with Attention-Based NLI Models

Gajbhiye, Amit; Winterbottom, Thomas; Al Moubayed, Noura; Bradley, Steven

Abstract

We consider the task of incorporating real-world commonsense knowledge into deep Natural Language Inference (NLI) models. Existing external knowledge incorporation methods are limited to lexical-level knowledge and lack generalization across NLI models, datasets, and commonsense knowledge sources. To address these issues, we propose a novel NLI model-independent neural framework, BiCAM. BiCAM incorporates real-world commonsense knowledge into NLI models. Combined with convolutional feature detectors and bilinear feature fusion, BiCAM provides a conceptually simple mechanism that generalizes well. Quantitative evaluations with two state-of-the-art NLI baselines on SNLI and SciTail datasets in conjunction with ConceptNet and Aristo Tuple KGs show that BiCAM considerably improves the accuracy the incorporated NLI baselines. For example, our BiECAM model, an instance of BiCAM, on the challenging SciTail dataset, improves the accuracy of incorporated baselines by 7.0% with ConceptNet, and 8.0% with Aristo Tuple KG.

Citation

Gajbhiye, A., Winterbottom, T., Al Moubayed, N., & Bradley, S. (2020). Bilinear Fusion of Commonsense Knowledge with Attention-Based NLI Models. In I. Farkaš, P. Masulli, & S. Wermter (Eds.), Artificial Neural Networks and Machine Learning – ICANN 2020 (633-646). Springer Verlag. https://doi.org/10.1007/978-3-030-61609-0_50

• Online Publication Date: Oct 22, 2020
• Publication Date: 2020
• Deposit Date: Oct 28, 2020
• Publicly Available Date: Oct 28, 2020
• Publisher: Springer Verlag
• Pages: 633-646
• Series Title: Lecture notes in computer science
• Series Number: 12396
• Book Title: Artificial Neural Networks and Machine Learning – ICANN 2020.
• ISBN: 9783030616083
• DOI: https://doi.org/10.1007/978-3-030-61609-0_50
• Public URL: https://durham-repository.worktribe.com/output/1626926
• Contract Date: Aug 13, 2020