12/1/2008 Overview • Discourse functions and devices, theory Augmenting Online Conversation applied through Automated Discourse • Spark, a messaging architecture Tagging Tagging • Video Vid • Testing of Spark’s advantages Student Presentation: Brynjar • Questions? Discourse functions and devices Discourse functions • Interactional, make sure that the • Discourse functions dictate the conversation goes on accordingly with intent/content of communication what the producers want • Discourse devices are the means to carry • Transactional, “the pool of shared , p out this intent out this intent knowledge” • Text chat lacks a lot of discourse devices • Grounding • Reinvent missing devices, e.g. with Spark • Mapping to a computable model: Discourse model, entities and structure Rhematic: I come in peace 1
12/1/2008 Discourse Devices Discourse devices, reloaded • Getting and acknowledging attention with • Text only chat is a recipe for visual cues misunderstanding, webcams suck • To transmit something rhematic, intonation • Avatars, virtual bodies to the rescue • Grounding with a nod G di ith d • Use good gaze control U d t l • Producing discourse structure transitions • How about automated nodding and with body language appropriate handwaving as well? Spark Markup example • Architecture to automatically diagnose and transform simple chat text to markup with discourse information • Discourse markup transformed to • Discourse markup transformed to behavioural markup • Behavioural markup interpreted to direct an avatar 2
12/1/2008 Video Spark components • Discourse module – POS with EngLite tagger – Finds lemma (root) of each word – Head phrase (noun or verb) and head phrase Head phrase (noun or verb) and head phrase modifier identification • Discourse context database • Avatar program Testing Questions? • Empirical • Objectively, time measured to finish task • Subjectively from questions answered by participants ti i t • Success! 3
Recommend
More recommend