Invited Speakers

Prof. James Pustejovsky

Department of Computer Science
Brandeis University
Waltham, USA

jamesp@cs.brandeis.edu

When Corpus Meets Theory: Creating Lexical Semantic Databases

Abstract: In this talk, I will present results of a recent approach to combining techniques in formal lexical semantics with methods in computational lexicography. This combination makes possible the design, construction, and maintenance of large lexical knowledge bases in a robust fashion. I will argue that blind statistical techniques, devoid of meaningful theoretical bias, are unable to form the basis for populating lexicons and lexical relations. On the other hand, the information derived from such techniques applied over large corpora can contribute in significant ways to the theoretical underpinnings of the language models adopted by the researcher. This work builds on the Corpus Pattern Analysis model of Pustejovsky and Hanks (2001), while significantly extending the model of Generative Lexicon Theory.

Prof. Misha Pavel

Department of Biomedical Engineering
Department of Electrical and Computer Engineering
Oregon Graduate Institute
Beaverton, USA

pavel@ece.ogi.edu

Fusion-Based Robust Signal Processing by Humans and Machines Using High-Dimensional Representation

Abstract: Although many existing automatic pattern recognition systems have achieved considerable success over the past fifty years, most of them lack robustness - the ability to perform as well as possible in unpredictable and changing environments. In contrast, biological system seem to be much more resilient to the environmental changes that are, at least partially, irrelevant to the tasks. The fact that current engineering systems lack the robustness found in natural systems leads to question some of the differences between the natural and the statistically optimal approaches. Our preliminary assessment of these differences led us to propose new methodology for pattern recognition, based on high-dimensional representation and information fusion.

In my presentation, I will note the wide range of benefits that can be derived from biological and machine data fusion, but I will then focus on fusion in service of pattern recognition. I will briefly describe a working hypothesis whereby data fusion in conjunction with neural-like computation and a high-dimensional representation can be used to achieve more robust pattern recognition performance that that obtained with more traditional approaches. I will illustrate the approach on one or two specific examples.

Prof. Misha Pavel's Biography

Dr. Pavel is a Professor and an Interim Department Head in the Department of Biomedical Engineering with a joint appointment in the Departments of Electrical and Computer Engineering and Computer Science and Engineering at OGI School of Science and Engineering at the Oregon Health Sciences University. His areas of research are at the intersection of engineering, biomedical signal processing, neuroscience, and cognitive science. At the intersection of computational neuroscience and cognitive science, Dr. Pavel is one of the leaders in the development of techniques for new approaches to Augmented Cognition. With the support of DARPA, Dr. Pavel is developing principled approaches to amplify human cognitive abilities that would enable operators to improve the quality and speed of their decision-making under uncertainty. The particular focus of his effort includes methodologies for allocation of attention and for representation of uncertainty.

His work in the engineering domains includes the development of algorithms for data fusion, intelligent signal processing, image fusion, pattern recognition, and speech processing. In close collaboration with NASA Ames Research Center and several industrial organizations, Dr. Pavel and his students have been developing sophisticated algorithms for image enhancement, registration and fusion, with emphasis on integration data from different sensors including millimeter wave radar, infrared imaging sensor, and low-light television camera. Information fusion is also at the heart of his efforts to develop robust pattern recognition techniques. As a member of the Center for Human and Computer Communication he participates in research and development of state of the art multimodal and perceptual user interfaces. His work on multimodal interfaces is based on his expertise in speech and image processing, pattern recognition, and cognitive science. During his leave of absence from OGI from1997 to 1999, Dr. Pavel was a Technology Leader and one of the founding members of AT&T Laboratories in Menlo Park where he was working on several applications involving networked applications including access to Internet via wireless, mobile devices. Prior to joining Oregon Graduate Institute in 1993, during his tenure as Associate Professor at New York University and earlier as an Assistant Professor at Stanford University, Dr. Pavel taught and conducted research in applications of cognitive psychology to the design of computer systems and interfaces. Prior to his academic career, Dr. Pavel was a member of the Technical Staff at Bell Laboratories, where his research included network analysis and modeling. Dr. Pavel is the author of more than 50 technical and scientific papers. He received his Ph.D. in Experimental Psychology from New York University; M.S. in Electrical Engineering from Stanford University, and B.S. in Electrical Engineering from Polytechnic Institute of Brooklyn.

Prof. Yorick Wilks

Department of Computer Science
University of Sheffield
Regent Court, 211 Portobello Street
Sheffield, England, S1 4DP

Y.Wilks@dcs.shef.ac.uk

METER: MEasuring TExt Reuse

Abstract: In this paper we present initial results from the METER (MEasuring TExt Reuse) project whose aim is to explore issues pertaining to text reuse and derivation, especially in the context of newspapers using newswire sources. Although the reuse of text by journalists has been studied in linguistics,

We are not aware of the investigation using existing computational methods for this particular task and context. In this paper we concentrate on classifying newspapers according to their dependency upon PA copy using a 3-class document-level scheme designed by domain experts from journalism and a number of well-known approaches to text analysis. We show that the 3-class document-level scheme is better implemented as 2 binary Naive Bayes classifiers and gives an F-measure score of 0.7309.

Prof. Ronald Allan Cole

Center for Spoken Language Research
University of Colorado
Boulder, USA

cole@cslr.colorado.edu

Perceptive Animated Interfaces: The Next Generation of Interactive Learning Tools

Abstract: We envision a new generation of human computer interfaces that engage users in natural face-to-face conversational interaction with intelligent animated characters. In specific learning domains, these perceptive animated interfaces will process auditory and visual signals presented by the user (e.g., speech sounds, lip movements, facial expressions, hand and body gestures) to interpret the user's spoken utterances and infer the user's intentions and cognitive state (e.g., focused, excited, frustrated). In addition, the system will use this information to build user models relevant to the learning task- e.g., if the system is helping the student learn about space science, it will analyze the student's responses to discover the student's preconceptions within that domain. Based on this information, the animated agent will orient to the user, provide real time feedback when the user speaks, and respond to the user much like a sensitive and effective teacher-through speech, facial expressions and hand and body gestures.

Research on perceptive animated interfaces at CSLR builds on advanced spoken dialogue systems research supported by DARPA and NSF. This research uses CU Communicator, an environment for researching and developing spoken dialogue systems that enable completely natural, unconstrained, mixed-initiative spoken dialogues in specific task domains. Communicator uses the public domain Galaxy hub-server architecture developed by the MIT speech group and maintained by MITRE under DARPA support. Spoken dialogue interaction in Communicator occurs via communication between users and various technology servers (all developed at CSLR) that pass messages through the Galaxy hub-audio server, speech recognizer, semantic parser, language generator, speech synthesizer, dialogue manager, and back-end servers that communicate with Web sites. By adding computer vision and computer animation servers, we have transformed Communicator into a platform for research and development of perceptive animated interfaces.

Our research on perceptive animated interfaces occurs in the context of Interactive Books: powerful learning tools that reside on client machines, and communicate with servers running Communicator. Interactive Books employ full-bodied 3D animated characters that integrate auditory and visual processing so the animated character can orient to the user, interpret the user's auditory and visual behaviors, and respond to these behaviors using speech, facial expressions and gestures. By studying the behaviors of master teachers working with individual students, and by working with these teachers to incorporate their best practices into our learning tools, we hope to invent animated agents that interact with students much like sensitive and effective teachers.

Our presentation will demonstrate the capabilities of Interactive Books, discuss the major research challenges involved, and describe ongoing work applying these learning tools to a number of learning domains, including foundational speech and reading skills, comprehension training and science education.

Prof. Ronald Allan Cole's Biography

University of Colorado, Boulder
Center for Spoken Language Research
MSSC-RL6
Campus Box 594
Phone: (303)- 735-5109
Email: cole@cslr.colorado.edu

Education

Ph.D., Psychology Univ. of California at Riverside, 1969-1971.
M.A., Psychology Univ. of California at Riverside, 1967-1969.
B.A., Psychology with Distinction, Univ. of Rochester, 1963-1967.

Research Interests and Accomplishments

Professor Ron Cole has studied speech recognition by human and machine for the past thirty years, and has published over 150 articles in scientific journals and published conference proceedings. In 1990, Ron founded the Center for Spoken Language Understanding (CSLU) at the Oregon Graduate Institute. In 1998, Ron founded the Center for Spoken Language Research (CSLR) at the University of Colorado, Boulder with Wayne Ward, John Hansen and Dan Jurafsky.

Ron's goals are to advance interactive language technologies and systems through basic research; to incorporate research advances into systems that facilitate learning and information access; and to make language resources accessible to all, so that students and educators worldwide can participate in developing new technologies and applications. To this end, Ron created the CSLU Toolkit while at OGI, a comprehensive set of free tools, tutorials and technologies enabling research and development of interactive language systems. The toolkit has been distributed to over 3000 sites in 62 countries and has become a widely used platform for research and application development in several languages. The tradition of providing free tools and technologies to researchers worldwide is being continued at CSLR, which offers its Conversational Agent Toolkit to create advanced dialogue systems, such as the CSLR Communicator system. Currently, Ron is working with colleagues at CU Boulder, the Boulder Valley School District, Oregon Graduate Institute, UC Santa Cruz, UC San Diego and the Universidad de las Americas in Puebla, Mexico to develop interactive reading tutors incorporating intelligent animated agents to help children learn to read. .

Ron is also a passionate advocate of international collaboration in computer science. He has organized five workshops designed to stimulate and sustain international cooperation to advance science and technology and solve global problems.

Professional Experience

Center Director/Professor, Center for Spoken Language Understanding, University of Colorado, Boulder, 10/98 - Present.

Professor, Department of Computer Science and Engineering, Oregon Graduate Institute, Beaverton, OR 4/92 - 10/98.

Associate Professor, Department of Computer Science and Engineering, Oregon Graduate Institute, Beaverton, OR 8/88 - 4/92.

Senior Project Scientist, Department of Computer Science, Carnegie Mellon University, 5/80 - 8/88.

Associate Professor, Department of Psychology, Carnegie Mellon University, 1/75 - 4/80.

Visiting Lecturer, Department of Linguistics, Tel-Aviv University, Tel-Aviv, Israel, 1/74 - 6/74.

Associate Professor, Department of Psychology, University of Waterloo, 7/74 - 12/74.
Assistant Professor, Department of Psychology, University of Waterloo, 7/70 - 6/74.

Selected Publications

Cole, R.A., Mariani, J., Uszkoreit, H., Varile Battista, G., Zaenen, A., Zampolli, A., Zue, V., (eds.), (1997). Survey of the state of the art in human language technology. Cambridge University Press and Giardini Editori.

Cole, R. A., Sutton, S., Yan, Y., Vermeulen, P., Fanty, M., "Accessible Technology for Interactive Systems: A New Approach to Spoken Language Research," ICASSP 1998. (Seattle, WA), pp. 1037 - 1040, May 1998.

Cole, R., Carmell, T., Connors, P., Macon, M., Wouters, J., de Villiers, J., Tarachow, A., Massaro, D., Cohen, M., Beskow, J., Yang, J., Meier, U., Waibel, A., Stone, P., Fortier, G., Davis, A., Soland, C., "Intelligent Animated Agents for Interactive Language Training," Proceedings of ESCA-StiLL, Marholmen, Sweden, May 1998.

Cole, R.A., D.G. Novick, P.J.E. Vermeulen, S. Sutton, M. Fanty, L.F.A. Wessels, J.H. de Villiers, J. Schalkwyk, B. Hansen, D. Burnett, "Experiments with a spoken dialogue system for taking the US Census," Speech Communication, Vol. 23, pp. 243 - 260, 1997.

Cole, R. A., L. Hirschman et al., "The Challenge of Spoken Language Systems: Research Directions for the Nineties," IEEE Transactions on Speech and Audio Processing, 3 (1), pp. 1 - 21, January, 1995.

Collaborators:

Wayne Ward, University of Colorado, Boulder
Victor Zue, MIT
Dan Jurafsky, University of Colorado, Boulder
Dom Massaro, University of California at Santa Cruz
Michael Cohen, University of California at Santa Cruz
John Hansen, University of Colorado, Boulder
Walter Kintsch, University of Colorado, Boulder
Javier Movellan, University of California, San Diego
Jan van Santen, Oregon Graduate Institute
Mike Macon, Oregon Graduate Institute
James Martin, University of Colorado, Boulder
Richard Olson, University of Colorado, Boulder
Lynn Snyder, University of Colorado, Boulder

TSD