Welcome to UIBK 2019
Welcome to the first workshop on User Interactions for Building Knowledge (UIBK), held in conjunction with the 24th ACM International Conference on Intelligent User Interfaces in Los Angeles, CA.
|09:20-10:05||Invited Talk: Towards Actionable Intelligence: Enhancing the Interplay of Data, Human and Machine
Xiaotong Liu, IBM Research Almaden
|10:05-10:30||WoS – Open Source Wizard of Oz for Speech Systems
Birgit Brüggemeier, Philip Lalone
|11:00-11:25||An apprenticeship model for human and AI collaborative essay grading
Alok Baikadi, Lee Becker, Jill Budden, Peter Foltz, Andrew Gorman, Scott Hellman, William Murray, Mark Rosenstein
|11:25-11:50||Detecting Changes in User Behavior to Understand Interaction Provenance during Visual Data Analysis
Alyssa Pena, Eric Ragan, Theodora Chaspari
|11:50-12:15||Eliciting Structures in Data
Ahmad Al-Shishtawy, Juhee Bae, Mohamed-Rafik Bouguelia, Göran Falkman, Sarunas Girdzijauskas, Olof Gönerup, Anders Holst, Alexander Karlsson, Slawomir Nowaczyk, Sepideh Pashami, Alan Said, Amira A. Soliman El Hosary
Fulfilling the promise of artificial intelligence (AI) - solving complex decision problems - depends on the ability to provide the underlying algorithms with domain specific information; data and knowledge that must come from human experts and in domains where needs, values, and often the validity of the information itself is continually shifting. Currently, building these models for AI systems to work from is an unintuitive task for users who have little to no knowledge of how the system and its underlying algorithms function. So in practice, this process is divided between AI-experts who understand what the system needs, and subject matter experts (SMEs) who understand the nuances of the data and domain. But this split is inefficient and often results in the stagnation of AI systems, because they cannot be updated quickly enough to keep up with the shifts in the real world situations where these systems are expected to operate. And in many complex problem spaces, designers and developers cannot completely define user needs; effective systems should have the ability for end users to tell the system what they need from it - eg. participate in training and knowledge capture.
Giving control over producing and refining these data sets and knowledge models directly to SMEs or even end users allows them ownership and agency over their data, knowledge, and AI needs. This workshop seeks to bring together researchers across different AI spaces to share and discuss the challenges of building interactions for guiding novice users through knowledge collection and model building or training.
Call for Papers
We are soliciting submissions that address any area of guiding novice users through training or model building, including but not limited to:
- curating the set of knowledge or training data (e.g., helping users understand what the system does and does not know)
- obtaining knowledge from users of various levels of expertise
- managing/mitigating human bias in the final system
- the role of explainability (XAI) during training/model building
- the effect of human generated knowledge in the explainability of the system’s decisions
- interactions for exchanging complex knowledge (fuzziness, multi-classification, etc.)
- refining systems from a pre-built model (maintenance, personalization)
- considerations around soliciting/collecting information from end users (ethics, error checking, data poisoning, etc)
- interaction modalities and novel visualizations for communicating model state to end users
We particularly encourage contributors to address and illustrate issues like these with case studies that explore the training/model building issues in specific subfields of AI such as training for machine learning classification, domain construction for automated planning, model refinement for data mining and more. This workshop seeks to bring together researchers and practitioners across different AI spaces to share and discuss challenges of and proposed solutions for building interactions for guiding novice users through knowledge collection and model building or training.
We invite submissions of two types: short papers and applied case studies.
Short papers (4-6 pages): presenting traditional research, work in progress, positions, emerging research issues, or lessons learnt that address challenges or questions affecting the design and implementation of user interactions for collecting and curating domain knowledge or training data. Papers may focus on a specific area of AI (eg. machine learning, planning, etc) or work that is applicable across different areas of AI or to AI systems where user interactions are agnostic to the underlying AI algorithms.
Applied case studies: may be demos, videos, or posters that discuss a particular challenge encountered during training of an AI-based system. All formats must include a 1-2 page written abstract. Actual demos, posters and videos may be submitted for consideration during review, but are not required until they are accepted for publication.
The papers and case studies will be juried by the organizers and program committee, then chosen according to relevance, quality, and interest to attendees. At least one author of each accepted submission must register for the workshop, and present its contents.
Paper Submissions Due:
December 14, 2018 December 21, 2018
Notification to Authors: January 14, 2019
Camera Ready Due: February 15, 2019
Workshop Date: March 20, 2019
Johanne Christensen, North Carolina State University (firstname.lastname@example.org)
Juhee Bae, University of Skövde (email@example.com)
Benjamin Watson, North Carolina State University (firstname.lastname@example.org)
Kartik Talamadupula, IBM Research (email@example.com)
Josef Spjut, NVIDIA (firstname.lastname@example.org)
Stacy Joines, IBM (email@example.com)