Generating Expository Dialogue from Monologue: Motivation, Corpus and Preliminary Rules

Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the ACL, pages 333–336, Los Angeles, California, June 2010. c⃝2010 Association for Computational Linguistics Generating Expository Dialogue from Monologue: Motivation, Corpus and Preliminary Rules Paul Piwek Centre for Research in Computing The Open University Walton Hall, Milton Keynes, UK p.piwek@open.ac.uk Svetlana Stoyanchev Centre for Research in Computing The Open University Walton Hall, Milton Keynes, UK s.stoyanchev@open.ac.uk Abstract Generating expository dialogue from mono- logue is a task that poses an interesting and re- warding challenge for Natural Language Pro- cessing. This short paper has three aims: firstly, to motivate the importance of this task, both in terms of the benefits of ex- pository dialogue as a way to present in- formation and in terms of potential applica- tions; secondly, to introduce a parallel cor- pus of monologues and dialogues which en- ables a data-driven approach to this challenge; and, finally, to describe work-in-progress on semi-automatic construction of Monologue- to-Dialogue (M2D) generation rules. 1 Introduction The tasks of text generation – e.g., Reiter et al. (2005) and Demir et al. (2008) – and generation in dialogue – e.g., Stent (2002) and DeVault et al. (2008) – are central topics in Natural Language Gen- eration (NLG). What sets the two tasks apart is the interactive nature of dialogue, where participants need to adapt their contributions to each other. This paper introduces an NLG task, the genera- tion of expository dialogue, to the Computational Linguistics community which occupies the middle ground between these two tasks. An expository di- alogue is an authored conversation between two fic- tive characters. It can be presented as text, audio or film. Although there is no real-time interactivity, in expository dialogue the contributions of the charac- ters do need to mesh with each other. The main pur- pose of expository dialogue is to present informa- tion (a description, explanation or definition) to the reader, hearer or viewer, in contrast with dramatic dialogue, which tells a story. The use of expository dialogue goes back as far as Plato (c. 470-399 BC), who expressed his ideas as dialogues between Socrates and his contemporaries. Recently, a number of empirical studies show that for some purposes expository dialogue has advan- tages over monologue: for learners, dialogue can be more memorable, stimulate them to formulate their own questions (Craig et al., 2000), and get them to talk with each other (Lee et al., 1998). Expository dialogue has also been found to be more effective for persuasion (Suzuki and Yamada, 2004). Additionally, dialogue lends itself very well for multimedia presentations by computer-animated agents (Andr´e et al., 2000; van Deemter et al., 2008). Potential application domains include ed- ucation, (serious) games and E-Health. In educa- tion, information from textbooks could be presented in dialogue form, possibly using virtual reality plat- forms such as Second Life. Automatically gener- ating dialogue from text for non-player characters could have a tremendous impact on the gaming in- dustry; e.g., (IGDA Game Writers SIG, 2003) state that the …

Content truncated to 500 words