The cost of reasoning with RDF updates

Abstract

Many real world RDF collections are large compared with other real world data structures. Such large RDF collections evolve in a distributed environment. Therefore, these changes between RDF versions need to be detected and computed in order to synchronize these changes to the other users. To cope with the evolving nature of the semantic web, it is important to understand the costs and benefits of the different change detection techniques. In this paper, we experimentally provide a detailed analysis of the overall process of RDF change detection techniques namely: explicit change detection, forward-inference change detection, backward-inference change detection and backward-inference and pruning change detection. The results show that pruning is relatively expensive by comparison with inferencing

Consistent RDF Updates with Correct Dense Deltas

“…This indicates that for the data structure used, the time required to carry out pruning exceeds the inference time both for ∆D c and ∆ED. This is consistent with previous findings [1]. The overall delta time shown in Figure 11 indicates that taking account of set difference operations, inferencing and pruning, approaches that prune the delta set tend to require significantly more processing power than non-pruning approaches.…”

“…The general rule for pruning is that if the subject or object of a triple in this triple cannot be inferred, consequently the triple can be pruned before the inference process begins. Although pruning may reduce the workload for inferencing, it carries a potential performance penalty [1].…”