Abstract
Complex reasoning over text requires understanding and chaining together free-form predicates 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 logical 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 combined with a reranker) on ROPES, a recently-introduced complex reasoning dataset.- Anthology ID:
- 2020.emnlp-main.245
- Volume:
- Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)
- Month:
- November
- Year:
- 2020
- Address:
- Online
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 3040–3050
- Language:
- URL:
- https://aclanthology.org/2020.emnlp-main.245
- DOI:
- 10.18653/v1/2020.emnlp-main.245
- Cite (ACL):
- Jiangming Liu, Matt Gardner, Shay B. Cohen, and Mirella Lapata. 2020. Multi-Step Inference for Reasoning Over Paragraphs. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 3040–3050, Online. Association for Computational Linguistics.
- Cite (Informal):
- Multi-Step Inference for Reasoning Over Paragraphs (Liu et al., EMNLP 2020)
- PDF:
- https://preview.aclanthology.org/auto-file-uploads/2020.emnlp-main.245.pdf
- Data
- ROPES