Towards a multi-way similarity join operator

Increasing volumes of data consumed and managed by enterprises demand effective and efficient data integration approaches. Additionally, the amount and variety of data sources impose further challenges for query engines. However, the majority of existing query engines rely on binary join-based query planners and execution methods with complexity that depends on the number of involved data sources. Moreover, traditional binary join operators are not able to distinguish between similar and different tuples, treating every incoming tuple as an independent object. Thus, if tuples are represented differently but refer to the same real-world entity, they are still considered as non-related objects. We propose MSimJoin, an approach towards a multi-way similarity join operator. MSimJoin accepts more than two inputs and is able to identify duplicates that correspond to similar entities from incoming tuples using Semantic Web technologies. Therefore, MSimJoin allows for the reduction of both the height of tree query plans and duplicated results.