A Dynamic Query Optimization on a Sparql Endpoint by Approximate Inference Processing

On a retrieval of Linked Open Data using SPARQL, it is important to construct an efficient query that considers its execution cost, especially when the query utilizes inference capability on the endpoint. A query often causes enormous consumption of endpoints´ computing resources since it is sometimes difficult to understand and predict what computations will occur on the endpoints. Preventing such an execution of time-consuming queries, approximating the original query could reduce loads of endpoints. In this paper, we present a preliminary idea and its concept on building endpoints having a mechanism to automatically avoid unwanted amount of inference computation by predicting its computational costs and allowing it to transform such a query into speed optimized query. Our preliminary experiment shows a potential benefit on speed optimizations of query executions by applying query rewriting approach. We also present a preliminary prototype system that classifies whether a query execution is time-consuming or not by using machine learning techniques at the endpoint-side.