An Execution Mechanism for Combining Query Packs and Once-Transformations

In Inductive Logic Programming (ILP), several techniques have been introduced to improve the efficiency of query execution. One such technique is query pack execution. A set of queries with a common prefix, as it is generated by the refinement operator of a typical ILP system, can be executed faster after it is converted into a tree structure called a query pack. On the other hand, query transformations improve the efficiency of executing a single query by transforming it into a different form that is more efficient to execute. Combining query packs with query transformations is difficult because a transformation may have a negative effect on the structure of the pack. The once-transformation is an important query transformation, and can improve the efficiency of query execution by several orders of magnitude. In this work, we extend query pack execution in such a way that it is able to handle queries produced by the oncetransformation. We do this in the context of ilProlog, a high performance Prolog system with specific extensions for supporting ILP systems. We evaluate our approach on both artificial domains and real world ILP applications. An Execution Mechanism for Combining Query Packs and Once-Transformations Remko Tronçon Henk Vandecasteele Jan Struyf Bart Demoen Gerda Janssens {remko,henkv,jan,bmd,gerda}@cs.kuleuven.ac.be Dept. of Computer Science, K.U. Leuven, Belgium Abstract In Inductive Logic Programming (ILP), several techniques have been introduced to improve the efficiency of query execution. One such technique is query pack execution. A set of queries with a common prefix, as it is generated by the refinement operator of a typical ILP system, can be executed faster after it is converted into a tree structure called a query pack. On the other hand, query transformations improve the efficiency of executing a single query by transforming it into a different form that is more efficient to execute. Combining query packs with query transformations is difficult because a transformation may have a negative effect on the structure of the pack. The once-transformation is an important query transformation, and can improve the efficiency of query execution by several orders of magnitude. In this work, we extend query pack execution in such a way that it is able to handle queries produced by the once-transformation. We do this in the context of ilProlog, a high performance Prolog system with specific extensions for supporting ILP systems. We evaluate our approach on both artificial domains and real world ILP applications.In Inductive Logic Programming (ILP), several techniques have been introduced to improve the efficiency of query execution. One such technique is query pack execution. A set of queries with a common prefix, as it is generated by the refinement operator of a typical ILP system, can be executed faster after it is converted into a tree structure called a query pack. On the other hand, query transformations improve the efficiency of executing a single query by transforming it into a different form that is more efficient to execute. Combining query packs with query transformations is difficult because a transformation may have a negative effect on the structure of the pack. The once-transformation is an important query transformation, and can improve the efficiency of query execution by several orders of magnitude. In this work, we extend query pack execution in such a way that it is able to handle queries produced by the once-transformation. We do this in the context of ilProlog, a high performance Prolog system with specific extensions for supporting ILP systems. We evaluate our approach on both artificial domains and real world ILP applications.