Efficient Evaluation of Queries in MystiQ Technical Report for Seminar on Probablistic Databases

We describe how approximate queries are useful and not well-supported by traditional systems. Probabilistic databases are a natural match for answering approximate queries. We outline similarity operators for use in approximate queries. Possible World Semantics are used as a reference model for query evaluation. Intensional semantics track the origin of complex probabilistic events with lineage. We describe how MystiQ addresses the problem of combinatoric data explosion with extensional semantics which only work for a subset of query plans. We show how to handle unsafe queries without a safe plan. Finally, we compare MystiQ to other probablistic database systems and discuss its limitations.