May 13, 2008
May 13, 2008
Location: Estoril, Portugal
Why Conversational Agents do what they do
The First Functional Markup Language Workshop
Following up on the papers presented at the Bologna workshop on Virtual Humans at AAMAS 2002, an international standardization effort started with the intention to specify a standard framework for the generation of multimodal communicative behavior and in particular the specification of interchange formats, or representation languages to be used as lingua franca between different modules within the conversational agent framework. So far, this initiative has resulted in a first specification of the so-called Behavioral Markup Language (BML) which serves as an interface between the behavior planning and the behavior realization modules within the generic SAIBA agent framework.
|Workshop at AAMAS 2008|
|May 13, 2008|
Also planned within this initiative is the specification of a Functional Markup Language (FML) which will serve as an interface language between the cognitive and the behavioral layer, i.e. between intent planning and behavior planning. FML is envisioned to describe communicative and expressive intent and provide - without any reference to concrete physical behavior - a description that accounts as much as possible for the aspects that are relevant and influential in the planning of verbal and nonverbal behavior. An FML description should thus fulfill two functions. First, it must define the basic semantic units associated with a communicative event. Second, it should allow the annotation of these units with properties that further describe communicative function such as expressive, affective, discursive, epistemic, or pragmatic functions.
Discussions about this representation language still need to start and gain momentum. They become more urgent now that more and more researchers commit themselves to the BML standards.
The aim of the workshop is to function as an incubator to initiate the discussion on FML, to learn about the opinions and ideas of the researchers on this topic, to learn about what functions are prominent in existing implementations, to get a good grip on the issues that need to be taken into account and the problems that need to be settled. The outcome should be a common understanding of what is involved in defining FML, first suggestions of what it should contain and especially an agenda to plan and execute the standardization work. The workshop will be of interest to the whole community of virtual human researchers that are concerned with specifying and implementing conversational agents. There is a growing need for options to exchange components as the complexity of agent behaviors is increasing which calls for a degree of standardization. Other groups within the agent community at large may also be interested in this effort to outline the contours of a general cognitive architecture for agents with emphasis on their communicative abilities.
The workshop is intended for discussion and will consist of the following sessions:
We invite position papers addressing one or more of the following issues:
- presentation by organizing committee summarizing position papers
- selected/short presentations
- discussion groups
- definition of research agenda
- planning of future actions
The main criteria for selection are:
- legacy: what functions are used/specified in current systems
- desires: what do researchers believe FML should specify
- framework: what are the implications for the SAIBA framework as a whole (http://wiki.mindmakers.org/projects:saiba:main/)
Papers should be between 4 to 8 pages. They should be send (.pdf)
to . There are no strict formatting guidelines for the first submission.
- clarity of ideas on how to progress
- quality and substance: are the ideas put forward solidly embedded in working practice or theory
Revised papers will be published in a book that is part of the CTIT Workshop Series.
|February 6, 2008: ||deadline for submission of papers|
|February 25, 2008: ||notification of acceptance|
|April 1, 2008:|| deadline for revised papers|
|May 12 or 13, 2008:|| workshop|
Dirk Heylen (University of Twente)
Stefan Kopp (University of Bielefeld)
Stacy Marsella (University of Southern California)
Catherine Pelachaud (Paris VIII University and INRIA)
Hannes Vilhjálmsson (Reykjavík University)
For more information contact
Dirk Heylen: .