Word Distributions for Thematic Segmentation in a Support Vector Machine Approach

Proceedings of the 10th Conference on Computational Natural Language Learning (CoNLL-X), pages 101–108, New York City, June 2006. c⃝2006 Association for Computational Linguistics Word Distributions for Thematic Segmentation in a Support Vector Machine Approach Maria Georgescul ISSCO/TIM, ETI University of Geneva 1211 Geneva, Switzerland maria.georgescul@eti.unige.ch Alexander Clark Department of Computer Science Royal Holloway University of London Egham, Surrey TW20 0EX, UK alexc@cs.rhul.ac.uk Susan Armstrong ISSCO/TIM, ETI University of Geneva 1211 Geneva, Switzerland susan.armstrong@issco.unige.ch Abstract We investigate the appropriateness of us- ing a technique based on support vector machines for identifying thematic struc- ture of text streams. The thematic seg- mentation task is modeled as a binary- classification problem, where the different classes correspond to the presence or the absence of a thematic boundary. Exper- iments are conducted with this approach by using features based on word distri- butions through text. We provide em- pirical evidence that our approach is ro- bust, by showing good performance on three different data sets. In particu- lar, substantial improvement is obtained over previously published results of word- distribution based systems when evalua- tion is done on a corpus of recorded and transcribed multi-party dialogs. 1 Introduction (Todd, 2005) distinguishes between “local-level top- ics (of sentences, utterances and short discourse seg- ments)” and “discourse topics (of more extended stretches of discourse)”.1 (Todd, 2005) points out that “discourse-level topics are one of the most elu- sive and intractable notions in semantics”. Despite this difficulty in giving a rigorous definition of dis- course topic, the task of discourse/dialogue segmen- tation into thematic episodes can be described by 1In this paper, we make use of the term topic or theme as referring to the discourse/dialogue topic. invoking an “intuitive notion of topic” (Brown and Yule, 1998). Thematic segmentation also relates to several notions such as speaker’s intention, topic flow and cohesion. In order to find out if thematic segment identi- fication is a feasible task, previous state-of-the-art works appeal to experiments, in which several hu- man subjects are asked to mark thematic segment boundaries based on their intuition and a minimal set of instructions. In this manner, previous studies, e.g. (Passonneau and Litman, 1993; Galley et al., 2003), obtained a level of inter-annotator agreement that is statistically significant. Automatic thematic segmentation (TS), i.e. the segmentation of a text stream into topically coher- ent segments, is an important component in ap- plications dealing with large document collections such as information retrieval and document brows- ing. Other tasks that could benefit from the thematic textual structure include anaphora resolution, auto- matic summarisation and discourse understanding. The work presented here tackles the problem of TS by adopting a supervised learning approach for capturing linear document structure of non- overlapping thematic episodes. A prerequisite for the input data to our system is that texts are divided into sentences or utterances.2 Each boundary be- tween two consecutive utterances is a potential the- matic segmentation point and therefore, we model the TS task as a binary-classification problem, where each utterance should be classified as …

Content truncated to 500 words