Unsupervised Sentence Textual Similarity with Compositional Phrase Semantics

Proceedings of the 29th International Conference on Computational Linguistics, pages 4976–4995 October 12–17, 2022. 4976 Unsupervised Sentence Textual Similarity with Compositional Phrase Semantics Zihao Wang Department of CSE HKUST Hong Kong, China zwanggc@cse.ust.hk Jiaheng Dou and Yong Zhang ∗ BNRist, RIIT, Institute of Internet Industry Department of Computer Science and Technology Tsinghua University, Beijing, China djh19@mails.tsinghua.edu.cn zhangyong05@tsinghua.edu.cn Abstract Measuring Sentence Textual Similarity (STS) is a classic task that can be applied to many downstream NLP applications such as text gen- eration and retrieval. In this paper, we focus on unsupervised STS that works on various domains but only requires minimal data and computational resources. Theoretically, we pro- pose a light-weighted Expectation-Correction (EC) formulation for STS computation. EC for- mulation unifies unsupervised STS approaches including the cosine similarity of Additively Composed (AC) sentence embeddings (Arora et al., 2017), Optimal Transport (OT) (Kus- ner et al., 2015), and Tree Kernels (TK) (Le et al., 2018). Moreover, we propose the Re- cursive Optimal Transport Similarity (ROTS) algorithm to capture the compositional phrase semantics by composing multiple recursive EC formulations. ROTS finishes in linear time and is faster than its predecessors. ROTS is empiri- cally more effective and scalable than previous approaches. Extensive experiments on 29 STS tasks under various settings show the clear ad- vantage of ROTS over existing approaches.1 Detailed ablation studies demonstrate the effec- tiveness of our approaches. 1 Introduction Sentence Textual Similarity (STS) measures the semantic equivalence between a pair of sentences, which is supposed to be consistent with human evaluation (Agirre et al., 2012). STS is also an ef- fective sentence-level semantic measure for many downstream tasks such as text generation and re- trieval (Wieting et al., 2019; Zhao et al., 2019; Nikolentzos et al., 2020; Çelikyilmaz et al., 2020). In this paper, we focus on unsupervised STS which is expected to compare texts of various domains but only requires minimal data and computational resources. ∗ Corresponding author. 1Our code can be found in https://github.com/ zihao-wang/rots. There are several typical ways to compute un- supervised STS, including 1) treat each sentence as an embedding by the Additive Composition (AC) (Arora et al., 2017) of word vectors, then estimate the STS of two sentences by their co- sine similarity; 2) treat each sentence as a prob- abilistic distribution of word vectors, then measure the distance between distributions. Notably, Opti- mal Transport (OT) (Peyré and Cuturi, 2019)2 is adopted to compute the STS (Kusner et al., 2015). OT-based approaches search for the best alignment with respect to the word-level semantics and result in state-of-the-art solution (Yokoi et al., 2020). In this paper, we argue that phrase-level seman- tics should also be exploited to fully understand the sentences. For example, “optimal transport” should be considered as a mathematical term rather than two independent words. Specifically, the phrase chunk is composed of lower-level chunks and is usually represented as a node in tree structures. The aforementioned AC and OT-based STS meth- ods are too shallow to include such structures. Tree Kernels (TK) (Le …

Content truncated to 500 words