ExplainabilityTransparencyAccountabilitycs.DB
Explaining Expert Search and Team Formation Systems with ExES
Kiarash Golzadeh, Lukasz Golab, Jaroslaw Szlichta
May 21, 2024
IEEE International Conference on Data Engineering
4 citations
Key Takeaways
-
Expert search systems lack transparency: State-of-the-art expert identification and team formation systems operate as "black boxes" on collaboration networks, making it unclear to users why certain individuals are recommended as experts for specific skill queries.
-
ExES provides two types of explanations: The tool uses factual explanations to show which skills and collaborations made someone identifiable as an expert, and counterfactual explanations to suggest concrete actions (new skills to develop or collaborations to pursue) that would increase someone's chances of being found as an expert.
-
Practical speed improvements enable real-world deployment: The researchers developed pruning strategies that make the explanation system up to 10 times faster than exhaustive search methods, making ExES feasible for interactive use without sacrificing the quality or conciseness of explanations provided to users.
-
Explanations are actionable for individuals: By showing both why someone was or wasn't selected and what specific changes could improve their visibility, the system empowers individuals to understand and potentially influence how they appear in expert search results.
Abstract
Expert search and team formation systems operate on collaboration networks, with nodes representing individuals, labeled with their skills, and edges denoting collaboration relationships. Given a keyword query corresponding to the desired skills, these systems identify experts that best match the query. However, state-of-the-art solutions to this problem lack transparency. To address this issue, we propose ExES, a tool designed to explain expert search and team formation systems using factual and counterfactual methods from the field of explainable artificial intelligence (XAI). ExES uses factual explanations to highlight important skills and collaborations, and counterfactual explanations to suggest new skills and collaborations to increase the likelihood of being identified as an expert. Towards a practical deployment as an interactive explanation tool, we present and experimentally evaluate a suite of pruning strategies to speed up the explanation search. In many cases, our pruning strategies make ExES an order of magnitude faster than exhaustive search, while still producing concise and actionable explanations.