PhD Dissertation Proposal by YOUNG Ezekiel Ong | Utility-Aware Behavioral Data Mining for Markets and Recommender Systems

ABOUT THE TALK

Behavioral data reveal preferences and choices, but prediction alone cannot identify decisions best serving a specified objective. This dissertation investigates utility-aware behavioral data mining for markets and recommender systems, linking behavioral evidence to seller revenue and profit in bundling and investor portfolio utility in financial recommendation.

The first study introduces the Integrated Price Mining Method (IPMM) for joint pricing of a predefined, potentially overlapping bundle catalog. Consumers may purchase multiple non-overlapping bundles. Modeling selection as weighted set packing and exploiting overlap dependencies, IPMM approximates revenue maximization. Evaluations show revenues near exact benchmarks on small instances and gains over baselines on larger ones.

The second study examines bundle construction and pricing by two competing firms with asymmetric inventories and heterogeneous costs. CoRe Bundling constructs differentiated offers using matching and rival-aware pricing; Strategic Replica reproduces and reprices feasible rival bundles. Evaluations identify conditions favoring differentiation or replication. Repeated responses reach exact equilibria in some small instances but often cycle, depending on initialization.

The third study develops GuidedRec to select a fixed number of behaviorally admissible non-held assets, maximizing estimated, profile-conditioned mean–variance utility of an equal-weight portfolio combining holdings and additions. Exact factor-state reformulation and safe dual completion bounds enable branch-and-bound search, certifying optimality over admissible slates or reporting valid gaps upon interruption. FAR-Trans evaluations show the highest observed date-balanced mean deployment-adjusted improvements in utility and Sharpe ratio among evaluated methods, with all main instances certified.

These studies develop computational methods for context-specific objectives amid bundle overlap, strategic competition, and asset covariance.

ABOUT THE SPEAKER

Ezekiel YOUNG is a PhD candidate at the School of Computing and Information Systems, Singapore Management University, under the supervision of Prof. Hady W. Lauw. His research focuses on the intersection of economics and computation, with applications to product bundling, competitive markets, and portfolio recommendation.