Corpus for the legal information processing system (CLIPS): A Chinese legal corpus annotated with discourse information

The paper presents the Corpus for the Legal Information Processing System, a corpus annotated with discourse relations based on Discourse Information Theory (DIT) that takes account of both macro- and micro-structures at discourse level. The information units and information elements in Chinese legal discourses are firstly characterized, and then a 16-valued classification of information units for macro-structure of discourse relations and a 25-valued classification of information elements for micro-structure of discourse relations are introduced. The paper also describes how the annotation strategy procedure is designed and the annotation conduction based on the above characterization.