Sentence Identification with BOS and EOS Label Combinations

Findings of the Association for Computational Linguistics: EACL 2023, pages 343–358 May 2-6, 2023 ©2023 Association for Computational Linguistics Sentence Identification with BOS and EOS Label Combinations Takuma Udagawa, Hiroshi Kanayama, Issei Yoshida IBM Research - Tokyo, Japan Takuma.Udagawa@ibm.com, {hkana,issei}@jp.ibm.com Abstract The sentence is a fundamental unit in many NLP applications. Sentence segmentation is widely used as the first preprocessing task, where an input text is split into consecutive sentences considering the end of the sentence (EOS) as their boundaries. This task formula- tion relies on a strong assumption that the input text consists only of sentences, or what we call the sentential units (SUs). However, real-world texts often contain non-sentential units (NSUs) such as metadata, sentence fragments, nonlin- guistic markers, etc. which are unreasonable or undesirable to be treated as a part of an SU. To tackle this issue, we formulate a novel task of sentence identification, where the goal is to identify SUs while excluding NSUs in a given text. To conduct sentence identification, we propose a simple yet effective method which combines the beginning of the sentence (BOS) and EOS labels to determine the most probable SUs and NSUs based on dynamic program- ming. To evaluate this task, we design an auto- matic, language-independent procedure to con- vert the Universal Dependencies corpora into sentence identification benchmarks. Finally, our experiments on the sentence identification task demonstrate that our proposed method gen- erally outperforms sentence segmentation base- lines which only utilize EOS labels. 1 Introduction The sentence, which we refer to as the sentential unit (SU), is a fundamental unit of processing in many NLP applications including syntactic pars- ing (Dozat and Manning, 2017), semantic parsing (Dozat and Manning, 2018), and machine transla- tion (Liu et al., 2020). Existing works mostly rely on sentence segmentation (a.k.a. sentence bound- ary detection) as the first preprocessing task, where we predict the end of the sentence (EOS) to split a text into consecutive SUs (Kiss and Strunk, 2006; Gillick, 2009). This approach relies on a strong assumption that the text only consists of SUs; how- ever, real-world texts like web contents often con- tain non-sentential units (NSUs) such as the meta- data of attachments embedded in the email body, repetition of symbols for separating texts, irregular series of nouns, etc. (just to name a few). Such NSUs may cause detrimental or unexpected results in the downstream tasks if considered as parts of the SUs and are more desirable to be distinguished from SUs in the first preprocessing step. To tackle this problem, we formulate a novel task of sentence identification, where the goal is to identify SUs while excluding NSUs in a given text (§3). This can be regarded as an SU span ex- traction task, where each SU span is represented by the beginning of the sentence (BOS) and the EOS labels.1 We illustrate the difference between sentence segmentation and sentence identification in Table 1. In sentence segmentation, the text frag- ment of an embedded file (“- TEXT.htm << File: TEXT.htm …

Content truncated to 500 words