Algorithmic Search for Fair and Successful Collaborative Teams
Standing
Undergraduate
Type of Proposal
Oral Research Presentation
Challenges Theme
Open Challenge
Faculty Sponsor
Dr. Hossein Fani
Proposal
Team formation aims to automate forming teams of experts who can successfully solve difficult tasks which have firsthand effects on creating an organizational performance. Forming a successful team is challenging due to the immense number of candidates with diverse backgrounds and personality traits. AI and Machine Learning facilitate analysis of massive collections of experts but largely ignore the fairness in recommended experts. Fairness breeds innovation and increases teams’ success by enabling a stronger sense of community, reducing conflict, and stimulating more creative thinking. However, there is little to no fairness-aware algorithmic method that considers fairness in team formation. The overarching theme of this work is to develop fast and fairness-aware team formation algorithms using AI-ML for recommending a fair list of experts in terms of sensitive attributes (e.g., popularity, age or gender), given required skills while almost promising the success of the recommended experts. We will address the fairness-aware team formation problem in four pipelined steps: 1) formally defining and quantifying fairness and utility, 2) designing novel AI-ML model structures or repurposing existing ones to obtain fairness, 3) employing data sampling heuristics and enumerating strategies to mitigate biases in the real-world data, and 4) fine-tuning the final ranked list of recommended experts. We will empirically benchmark our proposed method against state-of-the-art AI-ML team formation methods on three large-scale datasets namely computer science research groups (dblp), inventors (uspt), and movie industry (imdb) to demonstrate the consistency of our proposed methods across datasets from various domains.Â
Grand Challenges
Viable, Healthy and Safe Communities
Algorithmic Search for Fair and Successful Collaborative Teams
Team formation aims to automate forming teams of experts who can successfully solve difficult tasks which have firsthand effects on creating an organizational performance. Forming a successful team is challenging due to the immense number of candidates with diverse backgrounds and personality traits. AI and Machine Learning facilitate analysis of massive collections of experts but largely ignore the fairness in recommended experts. Fairness breeds innovation and increases teams’ success by enabling a stronger sense of community, reducing conflict, and stimulating more creative thinking. However, there is little to no fairness-aware algorithmic method that considers fairness in team formation. The overarching theme of this work is to develop fast and fairness-aware team formation algorithms using AI-ML for recommending a fair list of experts in terms of sensitive attributes (e.g., popularity, age or gender), given required skills while almost promising the success of the recommended experts. We will address the fairness-aware team formation problem in four pipelined steps: 1) formally defining and quantifying fairness and utility, 2) designing novel AI-ML model structures or repurposing existing ones to obtain fairness, 3) employing data sampling heuristics and enumerating strategies to mitigate biases in the real-world data, and 4) fine-tuning the final ranked list of recommended experts. We will empirically benchmark our proposed method against state-of-the-art AI-ML team formation methods on three large-scale datasets namely computer science research groups (dblp), inventors (uspt), and movie industry (imdb) to demonstrate the consistency of our proposed methods across datasets from various domains.Â