Compositional Language Understanding with Text-based Relational Reasoning
This work addresses a key challenge in language understanding for AI systems, though it is incremental as it builds on existing neural network methods with a new benchmark and baseline.
The authors tackled the problem of neural networks' limited ability to perform relational reasoning and combinatorial generalization in natural language by introducing a novel benchmark dataset that isolates these abilities, and they showed that a neural message-passing baseline with a relational inductive bias outperforms traditional recurrent neural networks in combinatorial generalization.
Neural networks for natural language reasoning have largely focused on extractive, fact-based question-answering (QA) and common-sense inference. However, it is also crucial to understand the extent to which neural networks can perform relational reasoning and combinatorial generalization from natural language---abilities that are often obscured by annotation artifacts and the dominance of language modeling in standard QA benchmarks. In this work, we present a novel benchmark dataset for language understanding that isolates performance on relational reasoning. We also present a neural message-passing baseline and show that this model, which incorporates a relational inductive bias, is superior at combinatorial generalization compared to a traditional recurrent neural network approach.