3rd Workshop on Affective and Attitude User Modeling

Assessing and Adapting to User Attitudes and Affect:
Why, When and How?

in conjunction with User Modeling 2003  

Pittsburgh, PA, USA
June 22, 2003

Background and Motivation
Focus Questions
Accepted Papers and Posters
Complete CFP

Please consult the
main conference website
for details regarding registration and accomodations.

All contributions will be made available on the Workshop Web site and published as part of an informal Annex to the main UM03 conference proceeding

Call For Papers

Background and Motivation:

User modeling has traditionally focused on what is generally considered ‘cognitive’ and ‘rational’ aspects of user behavior; typically the user’s knowledge and belief state. While useful, models focusing strictly on these aspects of user state often miss critical components of user mental state and behavior: affective states (e.g., basic affect such as like/dislike reactions; emotions such as frustration, fear, happiness, anger, etc.; moods), and attitudes (e.g. trust, doubt, etc.). These factors have also been referred to as ‘extrarational’ and shown to strongly influence both reasoning and communication. Over the past 5 years much progress has been made in a number of areas relevant to the assessment and modeling of these factors. This third workshop addressing affective and attitude user modeling issues follows the first and second workshops, held in Banff, BC, 1999 and Sonthofen, Germany (2001) .

As before, the main goal is to provide an opportunity for a focused exchange of ideas about this emerging subfield of user modeling. To this end, this 3rd one day workshop will address a variety of issues related to assessing user attitudes, affective states, and personality traits. The overall aim will be to explore core issues regarding the assessment, modeling, and adaptation to these states, across a range of applications (e.g., decision support systems, training and tutoring, telehealth and VR applications). These issues will include the following:

    Why is it important to assess these factors? What can be gained by augmenting current user models to include these factors?
    When should these factors be assessed (and adapted to), and when is it safe to ignore them? Are there situations when this type of assessment and adaptation might interfere with the human-machine interaction and the task at hand?
    What are the best methods available to accomplish this assessment, modeling, and adaptation? Can existing user modeling methods be adapted to include these factors, or must new methods and techniques be developed?
By addressing these issues in a mixed-mode, informal set of interactions, we hope to explore the feasibility and utility of attitude, affect, and personality user modeling, identify key problems to address, and contribute to advancing the state of the art of this emerging area of research.

Suggested Topics and Specific Questions to Be Addressed:

The workshop aims to address both theoretical and methodological issues, and applied issues in this broad research area. Examples of specific topics of interest include the core issues outlined above, and the following:

1: Methods for effective recognition of user emotional states and attitudes, with particular emphasis on multi-modal techniques.

2: Methods for modeling user emotional states and attitudes.

3: Methods and techniques for displaying and visualizing user emotional states and attitudes, and the use of these visualizations to promote interactive, iterative and accurate identification of these states in real-time.

4: Criteria for determining when user emotions and attitudes should be modeled and how this information should be used to enhance HCI.

5: Approaches to validating emotion and attitude in user models and their effectiveness in HCI.
Participants are encouraged to refer to the proceedings of the previous two workshops, and to build on existing results in the field. To assure focused interaction, participants should explicitly address one or more of the questions below in their submissions.

1: Taxonomy of Circumstances Requiring Affective and Attitude User Modeling
Can we begin to develop a taxonomy of circumstances under which affective/attitude user modeling is critical or useful, and identify a mapping between this taxonomy and the most appropriate level of modeling resolution?

2: Existing methods of Constructing Affective/Attitude User Models
What are the available means of constructing user affective/attitude models and what is the state of the art in applying these means to accurately identify user's affective states, attitudes, personality traits and associated behavioral patterns?

3: Validation and Evaluation
What are the best means of evaluating the effectiveness of an affective/attitude user model?

4: Guidelines for model use
What are the possible ways of using an affective/attitude user model, and how can we map these uses onto the taxonomy of situations where affect/attitude modeling is important ?

Workshop Format:

The exact workshop format and structure will be determined based on the submitted contributions. However, we will aim to provide a mix of interaction formats to stimulate discussion among participants, and to accommodate the presentation of both novel work-in-progress and more established state-of-the-art ideas and methods. To this end, the workshop will include a mix of the following:
    Traditional paper presentations with opportunities for questions and discussion

    Breakout sessions and presentation of session summaries to the entire workshop audience. (These sessions may be organized around the key questions listed above)

    Poster and demo sessions

    Keynote speaker

    One moderated panel discussion, with active encouragement of audience participation

    Concluding moderated panel discussion focusing on the key questions

Submission details and Contact Information:

The workshop organizers wellcome submissions of high-quality papers describing completed or on-going research addressing theories, methods, techniques and results in user emotion and attitude modeling. Both papers and posters may be submitted. Submission of interactive demonstrations is particularly encouraged.

All submissions should include an abstract, list of keywords and full author contact information. All submissions must be in English. Papers should not exceed 2000 words (excluding figures and tables). Authors should use the APA style of citations, two-column format and a font no smaller than 10 points. Submissions should be made in PDF or postscript formats.

Submissions should be sent electronically to evahud@earthlink.net and conati@cs.ubc.ca by March 10, 2003