Explainable User Models

Dear  Colleagues,

This special issue addresses research on Explainable User Models. As AI systems’ actions and decisions will significantly affect their users, it is important to be able to understand how an AI system represents its users. It is a well-known hurdle that many AI algorithms behave largely as black boxes. One key aim of explainability is, therefore, to make the inner workings of AI systems more accessible and transparent.

Such explanations can be helpful in the case when the system uses information about the user to develop a working representation of the user, and then uses this representation to adjust or inform system behavior. E.g., an educational system could detect whether students have a more internal or external locus of control, a music recommender system could adapt the music it is playing to the current mood of a user, or an aviation system could detect the visual memory capacity of its pilots. However,  when adapting to such user models it is crucial that these models are accurately detected. Furthermore, for such explanations to be useful, they need to be able to explain or justify their representations of users in a human-understandable way. This creates a necessity for techniques that will create models for the automatic generation of satisfactory explanations intelligible for human users interacting with the system.

The scope of the special issue includes but is not limited to:

Detection and Modelling

• Novel ways of Modeling User Preferences
• Types of information to model (Knowledge, Personality, Cognitive differences, etc.)
• Distinguishing between stationary versus transient user models (e.g., Personality vs Mood)
• Context modeling (e.g., at work versus at home, lean in versus lean out activities)
• User models from heterogeneous sources (e.g., behavior, ratings, and reviews)
• Enrichment and Crowdsourcing for Explainable User Models

Ethics

• Detection of sensitive or rarely reported attributes (e.g., gender, race, sexial orientation)
• Implicit user modeling versus explicit user modeling (e.g., questionnaires versus inference from behavior)
• User modeling for self actualization (e.g., user modeling to improve dietary or news consumption habits)

Human understandability

• Metrics and methodologies for evaluating fitness for the purpose of explanations
• Balancing completeness and understandability for complex user models
• Explanations to mitigate human biases (e.g., confirmation bias, anchoring)
• Effect of user model explanation on subsequent user interaction (e.g., simulations, and novel evaluation methodologies)

Effectiveness

• Analysis or comparison of context of use of explanation (e.g., risk, time pressure, error tolerance)
• Analysis of context of use of system (e.g., decision support, prediction)
• Analysis or comparison of effect of explaining in specific domains (e.g., education, health, recruitment, security)

Adaptive presentation of the explanations

• For different types of user
• Interactive explanations
• Investigation of which presentational aspects are beneficial to tailor in the explanation (e.g., level of detail, terminology, modality text or graphics, level of interaction)

Important Dates & Facts:
Manuscripts due by: February 20, 2022
Notification to authors: March 15, 2022

Prof. Dr. Nava Tintarev
Ms. Oana Inel
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Multimodal Technologies and Interaction is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 1600 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.