We present RUBISC, a new incremental chunker that can perform incremental slot filling and revising as it receives a stream of words. Slot values can influence each other via a unification mechanism. Chunks correspond to sense units, and end-of-sentence detection is done incrementally based on a notion of semantic/pragmatic completeness. One of RU-BISC’s main fields of application is in dialogue systems where it can contribute to responsiveness and hence naturalness, because it can provide a partial or complete semantics of an utterance while the speaker is still speaking. The chunker is evaluated on a German transcribed speech corpus and achieves a concept error rate of 43.3 % and an F-Score of 81.5. 1

I would like to express my sincere appreciation to A/P Lua Kim Teng, who patiently guided me through not only my PhD degree, but earlier on through my Honors and Master degrees. Without his help, guidance and counseling this thesis would definitely not have become a reality. My sincere gratitude as well to good friends like “Tat”, Hong I, Michelle and the “girls next door”, who not only kept me sane and centered through the ordeal of putting this thesis together, but also kept me well fed with cookies, “liang teh " and instant cereal through all those long hours of work. To them I owe all the weight that I’ve put on. To my students, who thoughtfully organized themselves when seeking help from me, to minimize the amount of time that I need to spend with them. To my family, who put up with my terrible tantrums, acid tongue and general crabbiness. Most of all to my beloved wife Catherine, who slaved for hours over stove and oven to bake me the cakes and cookies that kept me going through the night, and who was always willing to go to an empty bed as I spent night after night working on this

Humans understand spoken language in a continuous manner, incorporating complex semantic and contextual constraints at all levels of language processing on a word-by-word basis, but the standard paradigm for computational processing of language remains sentence-at-a-time, and does not demonstrate the tight integration of interpretations at various levels of processing that humans do. We introduce the fruit carts task domain, which has been specifically designed to elicit language that requires this sort of continuous understanding. A system architecture that incrementally incorporates feedback from a real-world reference resolution module into the parser is presented as a major step towards a continuous understanding system. A preliminary proof in principle shows that real-world knowledge can help resolve certain parsing ambiguities, thus improving accuracy, and that the efficiency of the parser, as measured by the number of constituents built, improves by upwards of 30 % on certain example sentences with multiple attachment ambiguities. A 26 % efficiency improvement was achieved for a dialogue transcript taken from those collected for the fruit carts task domain. We also argue that real-world reference information can help resolve ambiguities in speech recognition. Continuous Understanding of Spoken Language There are a number of speech-to-intention dialogue systems which undertake the task of understanding and/or interperting spoken language, such as Verbmobil (Kasper et al. 1996;

Contents 1 Introduction: Conversational Agents 1 2 Continuous Understanding 2 2.1 The Incremental Processing Alternative . . . . . . . . . . . . . . 3 2.2 People Understand Continuously . . . . . . . . . . . . . . . . . . 3 2.3 Weak vs. Strong AI in Human-Computer-Interaction . . . . . . . 7 2.4 Practical Value of Continuous Understanding . . . . . . . . . . . 7 2.5 Some Nascent attempts at Continuous Understanding . . . . . . 9 3 Intention Recognition Reigns Supreme 10 3.1 As a Probability Maximization . . . . . . . . . . . . . . . . . . . 13 3.2 What If the Intention is Wrong? . . . . . . . . . . . . . . . . . . 15 3.3 A Productive Interaction . . . . . . . . . . . . . . . . . . . . . . . 16 4 CAFE: In Search of an Architecture 16 4.1 Traditional Architectures . . . . . . . . . . . . . . . . . . . . . . 16 4.2 Some Non-Traditional Alternatives . . . . . . . . . . . . . . . . . 17 4.2.1 Giant BlackBoard System . . . . . . . . . . . . . . .

Short abstract Contents 1 Introduction 2 1.1 Motivations and Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2 Empiric evidence for Brittleness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.4 Defining Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.5 Robustness at different Linguistic Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.5.1 Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.5.2 Syntax Analysis and Parsing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.6 Semantics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.7 Pragmatics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.8 Evaluation . . . . . . . . . . . . . . . . . . . . . . ...

In [1] and [2] Tan and Lua introduce the concept of Belief Augmented Frames (BAF). In this paper we explore how BAFs may be used to model discourse, including how to model fuzzy linguistic hedges like "very" and "quite". We also demonstrate BAF reasoning with a simple toy example, and suggest topics for further research.

LOGUS is a French-speaking spoken language understanding (SLU) system which carries out a deeper analysis than those achieved by standard concept spotters. It is designed for multi-domain conversational systems or for systems that are working on complex application domains. Based on a logical approach, the system adapts the ideas of incremental robust parsing to the issue of SLU. The paper provides a detailed description of the system as well as results from two evaluation campaigns that concerned all of current French-speaking SLU systems. The observed error rates suggest that our logical approach can stand comparison with concept spotters on restricted application domains, but also that its behaviour is promising for larger domains. The question of the generality of the approach is precisely addressed by our current investigations on a new task: SLU for an emotional robot companion for young hospital patents. 1