Learning the Fine-Grained Information Status of Discourse Entities

Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics, pages 798–807, Avignon, France, April 23 - 27 2012. c⃝2012 Association for Computational Linguistics Learning the Fine-Grained Information Status of Discourse Entities Altaf Rahman and Vincent Ng Human Language Technology Research Institute University of Texas at Dallas Richardson, TX 75083-0688 {altaf,vince}@hlt.utdallas.edu Abstract While information status (IS) plays a cru- cial role in discourse processing, there have only been a handful of attempts to automat- ically determine the IS of discourse entities. We examine a related but more challenging task, fine-grained IS determination, which involves classifying a discourse entity as one of 16 IS subtypes. We investigate the use of rich knowledge sources for this task in combination with a rule-based approach and a learning-based approach. In experi- ments with a set of Switchboard dialogues, the learning-based approach achieves an ac- curacy of 78.7%, outperforming the rule- based approach by 21.3%. 1 Introduction A linguistic notion central to discourse processing is information status (IS). It describes the extent to which a discourse entity, which is typically re- ferred to by noun phrases (NPs) in a dialogue, is available to the hearer. Different definitions of IS have been proposed over the years. In this paper, we adopt Nissim et al.’s (2004) proposal, since it is primarily built upon Prince’s (1992) and Eck- ert and Strube’s (2001) well-known definitions, and is empirically shown by Nissim et al. to yield an annotation scheme for IS in dialogue that has good reproducibility.1 Specifically, Nissim et al. (2004) adopt a three- way classification scheme for IS, defining a dis- course entity as (1) old to the hearer if it is known to the hearer and has previously been referred to in the dialogue; (2) new if it is unknown to her and 1It is worth noting that several IS annotation schemes have been proposed more recently. See G¨otze et al. (2007) and Riester et al. (2010) for details. has not been previously referred to; and (3) me- diated (henceforth med) if it is newly mentioned in the dialogue but she can infer its identity from a previously-mentioned entity. To capture finer- grained distinctions for IS, Nissim et al. allow an old or med entity to have a subtype, which subcat- egorizes an old or med entity. For instance, a med entity has the subtype set if the NP that refers to it is in a set-subset relation with its antecedent. IS plays a crucial role in discourse processing: it provides an indication of how a discourse model should be updated as a dialogue is processed in- crementally. Its importance can be reflected in part in the amount of attention it has received in theoretical linguistics over the years (e.g., Halli- day (1976), Prince (1981), Hajiˇcov´a (1984), Vall- duv´ı (1992), Steedman (2000)), and in part in the benefits it can potentially bring to NLP applica- tions. One task that could benefit from knowledge of IS is identity coreference: since new entities by …

Content truncated to 500 words