ResearchPreference Data Management
Preference data is prevalent variety of domains. The past decade has seen a proliferation in the volume of preference data, the diversity of applications based on preference data, and the richness of preference analysis methods. Examples include rank aggregation in genomic data, management and analysis of elections, and recommendation systems in e-commerce. Management of preference data entails special requirements and opportunities that cast it substantially distinct from general (e.g., relational or graph) data management: certain atomic operations are not supported by standard database engines, yet are very common in preference data analytics; statistical models of preferences fall outside the standard formalisms of “probabilistic databases;” election data requires out-of-databases analysis of the space of outcomes for a given incomplete picture of the voter preferences. Our research develops fundamental frameworks for the management of preference data, with emphasis on the associated complexity and algorithms.