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.

翻译：近年来，基于Transformer的嵌入方法在句子表征领域占据主导地位。尽管它们在自然语言处理任务（如语义文本相似性（STS）任务）上取得了显著性能，但其黑盒性质和大数据驱动的训练方式引发了包括偏见、可信度与安全性在内的诸多担忧。已有许多工作致力于提升嵌入模型的可解释性，但这些问题尚未得到根本性解决。为实现内在可解释性，我们提出了一种纯粹白盒化且类人的句子表征网络——PropNet。受认知科学发现的启发，PropNet基于句子所包含的命题构建了一个层次化网络。实验表明，尽管PropNet在STS任务上与最先进的嵌入模型相比存在明显差距，案例分析揭示了其巨大的改进空间。此外，PropNet使我们能够分析和理解STS基准测试背后的人类认知过程。