AAAI 2004 Fall Symposium Series

STYLE AND MEANING IN LANGUAGE, ART, MUSIC
AND DESIGN

October 21-24 in Washington, D.C

In recent years a growing number of researchers working in artificial intelligence, cognitive science, computer graphics, computer music, and multimedia have begun to explicitly address issues of 'style' or connotative semantics in their work.  While it is still difficult to precisely characterize these concepts satisfactorily (we know it when we see it), common denominators of much of this work are: an emphasis on manner rather than topic, a focus on affective aspects of expression and understanding, and a search for 'dense' representations of meaning in which elements simultaneously symbolize multiple layers of meaning at once.

Recent areas of research in this vein have included forensic authorship attribution, information retrieval based on document genre or affect, composition of new music in a given composer's style, rendering animation in different motion styles, analyzing architectural styles for function and affect, and much more. Work in all media shares the problem of formalizing a notion of style, and developing a modeling language that supports the representation of differing styles.  However, due to the widely varying technical requirements of work in different media, little communication has traditionally existed between different 'style researchers'. 

The symposium will provide a unique meeting ground for researchers and practitioners in all media that share the problem of formalizing a notion of style, being an effective means to generate a discourse across diverse forms and approaches.  The goal of this symposium is to bring such individuals together, to seek out common languages and frameworks for discussion, as well as to establish a shared set of stylistic tasks, which can be used as a test-bed for extending and generalizing stylistic work. Keying into stylistic sensibility, we hope to make headway into understanding style in an attempt to develop methodology and modeling language for representation, analysis and generation of differing styles across multiple domains.

THE CHALLENGE

While much work remains in developing shared formalisms for research on style and connotation, we outlined a set of questions, which are more-or-less common to work in all various media.  These "challenge questions" will serve as foci for the symposium:

· Is there a general theory for style, which cuts across all kinds of human intellectual behavior? What is the relation between style and other content (e.g. informational) in the work you will be reporting at the symposium?

· Is there a general theoretical structure for the context that informs style and connotation that can be applied usefully in disparate media? Are there lessons in work you will be reporting at the symposium that are generalizable across media and genre?

· In operational terms, what are useful models and effective algorithms of the process of learning and producing style, and how can such models inform our understanding of stylistic features in the resulting work? In the work you will be reporting at the symposium - can the models and algorithms be used for both understanding style and generating style?

· Is style at the forefront of people's understanding the medium and discourse in the community you have worked with? How is style explicitly discussed or implicitly understood? How are stylistic distinctions learnt and transmitted to others within the community of recipients? In the work you are presenting, how is style understood by the intended audience?

· How can we usefully model the social context of a work, as a resource for understanding its style, its meaning, and its effect? Does the work you report take the context and effect of style outside the medium itself into account?

· What are the processes affecting stylistic diffusion among members of a discourse community?  What properties of the social context may affect the transmission or evolution of distinctive styles?  How is the work you are presenting affected by understanding the social networks in which style is embedded?

THE SYMPOSIUM VENUE

The symposium will be held Friday through Sunday, October 22-24. A one-day AI funding seminar, which will be open to all registered attendees of the fall symposium series, will precede the Symposium Series on Thursday, October 21.

To facilitate interaction among participants, the symposium will include, in addition to a number of "traditional" research paper presentations: (a) invited talks by leading figures in style research, )b) brief tutorials on work in individual media, and (c) panel discussions discussing various practical issues in research on style.

Research talks at the symposium include papers and demonstrations of works in domains as widely varying as text, poetry, caricature, game-playing,
theater and media art, architecture, and design.  Sessions each day will be preceded by invited talks and demonstrations by leading experts in applying
computational techniques to style in various media and domains.  Discussion panels will be held at the end of each day, focusing on fundamental questions for the development of a research community devoted to "computational approaches to style".

SCHEDULE

Click on the link to see the schedule.

The schedule might be subject to last minute changes.

PAPERS

Working notes containing most research papers is available here.

Full technical report will be published by AAAI.

REGISTRATION AND HOTELS

If you have not yet registered for the meeting (and intend to do so), please do so soon (at http://www.aaai.org/Symposia/Fall/2004/, as reduced hotel

rates are available only until September 27.

Invited Speakers:
Eduard Hovy, ISI (Natural Language)
Christopher Raphael,  (Computer Music)
George Stiny, MIT (Architecture)
Harold Cohen (Visual Arts)

Additional Information

For additional information about AAAI and the Symposium Series please see http://www.aaai.org/Symposia/symposia.html

ORGANIZING COMMITTEE

Shlomo Dubnov, University of California San Diego, USA (chair)

Shlomo Argamon, Illinois Institute of Technology, USA (co-chair)

Julie Jupp,The University of Sydney, Australia (co-chair)

Roger Dannenberg, Carnegie Mellon, USA

Graeme Hirst, University of Toronto, Canada

Jussi Karlgren, Swedish Institute of Computer Science, Sweden

Moshe Koppel, Bar-Ilan University, Israel

Rivka Oxman, Technion, Israel

Mine Ozkar, Massachusetts Institute of Technology, USA

James Shanahan, Clairvoyance Corporation, USA