Supporting Algorithm Accountability using Provenance — Opportunities and Challenges is a half-day satellite event that is to take place as part of Provenance Week 2018 at King’s College London, 12 July, 2018. The workshop aims to bring together researchers from different research areas, including law, machine learning, ethics as well as provenance, to explore challenges and opportunities for provenance-based algorithm accountability.
Concerns about the fairness of computational algorithms are increasingly growing. It is widely recognised that algorithms are playing an important role in our life. And more of us are demanding more transparency regarding how our personal data might be processed and used by algorithms. This demand for transparency and accountability in algorithms is not new. It has been an active research topic in areas like Artificial Intelligence, Human Computer Interactions and etc.
Provenance information has a long standing history in supporting scientific research and enhancing its rigour. Validating and enabling reproducibility of research results are part of the core motivations for provenance research. However, the potential role of provenance information in facilitating algorithm transparency and accountability is less known or well understood. Therefore, this half-day workshop aims to bring together researchers from different research areas, including law, machine learning, ethics as well as provenance, to explore challenges and opportunities for provenance-based algorithm accountability.
The workshop has three primary goals:
- To solicit and discuss case studies that will benefit from provenance research and technologies for supporting algorithm accountability
- To facilitate research collaborations and initiatives between provenance researchers and researchers from other areas working on algorithm accountability
- To identify gaps and open challenges in provenance research to support algorithm accountability
- Jun Zhao, University of Oxford
- Reuben Binns, University of Oxford
- Adriane Chapman, University of Southampton
Our workshop will be structured as follows:
- Invited talks by experts on algorithm accountability from other areas, including legal, ethical, human computer interactions and machine learning ~1 hour
- Short presentations of reflection or work-in-progress papers, ~1 hour
- Panel-led discussions about open challenges and opportunities for provenance-based algorithm explanations ~0.5 hour
Confirmed invited speakers include:
- Reuben Binns, Department of Computer Science, University of Oxford
- Paolo Missier, School of Computing, Newcastle University
- Sophie Stalla-Bourdillon, Southampton Law School, University of Southampton
- Jat Singh, Department of Computer Science & Technology, University of Cambridge
- Michael Veale, Department of Science, Technology, Engineering and Public Policy, University College London
Topics of interest
We invite short talks from the following themes:
- Analysing, querying, and reasoning about provenance for supporting algorithm accountability
- Case studies of using provenance to support algorithm accountability
- User studies of making provenance information more understandable for end users who look for more transparency and accountability
- Overviews of active projects working on provenance and algorithm accountability
- Reviews of provenance techniques in support of algorithm accountability
- Methods and data models for capturing provenance
- Position papers that review challenges and opportunities of supporting algorithm accountability using provenance
Authors are expected to first submit a 1-page expression of interest (up to 500 words). Full papers are invited based on quality of work and relevance to the workshop.
Both the Expression of Interest and the full paper should be submitted by emailing jun.zhao @ cs.ox.ac.uk
- Expression of interest: Tuesday, May 29
- Invitation to full paper: Friday, June 1
- Full paper submission: Monday, July 2
- Workshop: 1:30pm-5pm, July 12
All deadlines set at midnight Hawaii Standard Time.