SICSA XAI 2021
SICSA XAI Workshop 2021

Table of Contents

• Preface
  Summary: There were 13 papers submitted for peer-review to this workshop. Of these submissions, 12 papers were accepted for presentation at the workshop and 11 papers are presented in this volume. During the workshop, paper presentations were recorded with authors permission and are available at the following link.

Session 1: Applying and Evaluating Explanations

• Counterfactual Explanations for Student Outcome Prediction with Moodle Footprints 1 - 8
  Anjana Wijekoon, Nirmalie Wiratunga, Ikechukwu Nkisi-Orji, Kyle Martin, Chamath Palihawadana, David Corsar
• Are Contrastive Explanations Useful? 9 - 16
  James Forrest, Somayajulu Sripada, Wei Pang, George M. Coghill
• I don't understand! Evaluation Methods for Natural Language Explanations 17 - 24
  Miruna Clinciu, Arash Eshghi, Helen Hastie

Session 2: Roles within an XAI System and Accountability

• Towards Accountability Driven Development for Machine Learning Systems 25 - 32
  Chiu Pang Fung, Wei Pang, Iman Naja, Milan Markovic, Peter Edwards
• On Evidence Capture for Accountable AI Systems 33 - 39
  Wei Pang, Milan Markovic, Iman Naja, Chiu Pang Fung, Peter Edwards
• Challenges and Future Directions for Accountable Machine Learning 40 - 47
  Agne Zainyte, Wei Pang
• Human-in-the-Loop Approach to Digitisation of Engineering Drawings 48 - 55
  Andrew M. Fagan, Graeme M. West, Stephen D. J. McArthur

Session 3: Searching for Explanations

• Explaining Pen and Paper Puzzles with MUSes? 56 - 63
  Joan Espasa, Ian P. Gent, Ruth Hoffmann, Christopher Jefferson, Matthew J. McIlree, Alice M. Lynch
• Mining Markov Network Surrogates to Explain the Results of Metaheuristic Optimisation 64 - 70
  Alexander E. I. Brownlee, Aidan Wallace, David Cairns

Session 4: Position Papers

• Use of Explainable AI to Refine Artificial Immune System Algorithms 71 - 74
  Rachana R. Patel
• Non-Deterministic Solvers and Explainable AI through Trajectory Mining 75 - 78
  Martin Fyvie, John A. W. McCall, Lee A. Christie