Understanding Roles and Entities: Datasets and Models for Natural Language Inference
This work addresses a specific issue in NLI for AI researchers, offering incremental improvements through new datasets and a modified attention mechanism.
The authors tackled the problem of neural Natural Language Inference models failing to capture entities and roles, leading to errors like inferring 'Peter signed a deal' from 'John signed a deal'. They introduced two new datasets and a novel attention mechanism, resulting in models that perform as well as existing ones on standard benchmarks and significantly better on the new benchmarks for roles and entities.
We present two new datasets and a novel attention mechanism for Natural Language Inference (NLI). Existing neural NLI models, even though when trained on existing large datasets, do not capture the notion of entity and role well and often end up making mistakes such as "Peter signed a deal" can be inferred from "John signed a deal". The two datasets have been developed to mitigate such issues and make the systems better at understanding the notion of "entities" and "roles". After training the existing architectures on the new dataset we observe that the existing architectures does not perform well on one of the new benchmark. We then propose a modification to the "word-to-word" attention function which has been uniformly reused across several popular NLI architectures. The resulting architectures perform as well as their unmodified counterparts on the existing benchmarks and perform significantly well on the new benchmark for "roles" and "entities".