A Corpus of Preposition Supersenses in English Web Reviews

本研究介绍了首个带有介词超感标注的语料库，该语料库由爱丁堡大学等机构创建，专注于英语网络评论中介词的语义功能分类。数据集包含4250个介词标记，通过层次化组织的不词汇化类别进行全面手动标注。数据集的创建过程涉及对英语介词标记在网络评论语料库中的全面标注，并根据标注过程中遇到的数据调整超感层次结构。该数据集旨在解决介词语义歧义问题，支持下游任务如翻译或语义解析，为介词感消歧系统的开发提供基础。

This study introduces the first corpus annotated with preposition supersenses, developed by institutions including the University of Edinburgh, which focuses on classifying the semantic functions of prepositions in English online reviews. The dataset consists of 4250 preposition tokens, and is comprehensively manually annotated using hierarchically organized delexicalized categories. The construction of the dataset entails comprehensive annotation of English preposition tokens within an online review corpus, with the supersense hierarchy being adjusted in accordance with the data encountered during the annotation process. This dataset is designed to resolve prepositional semantic ambiguity, support downstream tasks such as machine translation and semantic parsing, and lay a foundation for the development of preposition sense disambiguation systems.