PropNet: a White-Box and Human-Like Network for Sentence Representation
This addresses interpretability issues in NLP for researchers and practitioners, though it is incremental as it builds on existing cognitive science insights without achieving SOTA performance.
The authors tackled the problem of black-box sentence embeddings by proposing PropNet, a white-box, human-like network based on propositions, but it showed a significant performance gap compared to state-of-the-art models in semantic textual similarity tasks.
Transformer-based embedding methods have dominated the field of sentence representation in recent years. Although they have achieved remarkable performance on NLP missions, such as semantic textual similarity (STS) tasks, their black-box nature and large-data-driven training style have raised concerns, including issues related to bias, trust, and safety. Many efforts have been made to improve the interpretability of embedding models, but these problems have not been fundamentally resolved. To achieve inherent interpretability, we propose a purely white-box and human-like sentence representation network, PropNet. Inspired by findings from cognitive science, PropNet constructs a hierarchical network based on the propositions contained in a sentence. While experiments indicate that PropNet has a significant gap compared to state-of-the-art (SOTA) embedding models in STS tasks, case studies reveal substantial room for improvement. Additionally, PropNet enables us to analyze and understand the human cognitive processes underlying STS benchmarks.