Reconciling Multiple Categorical Preferences with Double Pareto-based Aggregation Full text

Nikos Bikakis, Karim Benouaret, Dimitris Sacharidis
19th International Conference on Database Systems for Advanced Applications (DASFAA '14)
Abstract. Given a set of objects and a set of user preferences, both defined over a set of categorical attributes, the Multiple Categorical Preferences (MCP) problem is to determine the objects that are considered preferable by all users. In a naıve interpretation of MCP, matching degrees between objects and users are aggregated into a single score which ranks objects. Such an approach, though, obscures and blurs individual preferences, and can be unfair, favoring users with precise preferences and objects with detailed descriptions. Instead, we propose an objective and fair interpretation of the MCP problem, based on two Pareto-based aggregations. We introduce an efficient approach that is based on a transformation of the categorical attribute values and an index structure. Moreover, we propose an extension for controlling the number of returned objects. An experimental study on real and synthetic data finds that our index-based technique is an order of magnitude faster than a baseline approach, scaling up to millions of objects and thousands of users.