Information integration systems provide a uniform query interface to a set of sources. One of the key challenges for an information integration system is to provide maximally complete answers to user queries and to execute user queries efficiently. We describe an approach to map recursive datalog programs into a streaming, dataflow execution system. We show that our method can be used in conjunction with the Inverse Rules algorithm to create a new information integration system that can provide maximally complete answers to user queries and efficiently execute those queries. Our preliminary results show that in addition to generating maximally complete answers, we obtain performance improvements ranging from 8% to 24.3% over datalog execution.
Snehal Thakkar, Craig A. Knoblock