作者: Eleni Miltsakaki , Livio Robaldo , Alan Lee , Aravind Joshi
DOI: 10.1007/978-3-540-78135-6_23
关键词: Linguistics 、 Annotation 、 Argument (linguistics) 、 Computer science 、 Lexical item 、 Semantics 、 Attribution 、 Artificial intelligence 、 Discourse relation 、 Discourse connectives 、 Natural language processing 、 Treebank 、 Structure (mathematical logic)
摘要: An important aspect of discourse understanding and generation involves the recognition processing relations. These are conveyed by connectives, i.e., lexical items like because as a result or implicit connectives expressing an inferred relation. The Penn Discourse TreeBank (PDTB) provides annotations argument structure, attribution semantics connectives. In this paper, we provide rationale tagset, detailed descriptions senses with corpus examples, simple semantic definitions each type sense tags well informal inferences allowed at level.