Querying of several DL-Lite knowledge bases from various information sources-based polynomial response unification approach

Abstract

The inconsistency in the hierarchical knowledge base is due to the fact that the assertions (ABoxes) come from several sources with different levels of reliability. We introduce the management of this inconsistency problem to query the inconsistent DL-Lite knowledge bases. Most of the approaches studied in the literature first repair all of the inconsistent assertions in the DL-Lite incoherent knowledge base and then question it. In this article, we proceed directly to a knowledge base query to retrieve an exhaustive list of responses and then, once these responses are retrieved, we repair them. The main contribution of this work therefore consists of studying several response strategies to queries from various data sources using a recurrent function to calculate the rank of coherence which will allow us to manage inconsistencies in the set of responses. We study these strategies from different points of view: productivity, precision, recall, F-measure, Accuracy, G-mean and computational complexity. The experiments carried out, on different corpora, and the analysis of the results, which we carried out, show encouraging results in productivity since our approach is much more productive, providing the greatest number of responses, while remaining the best in time compared to the other studied approaches.