Learning DLP from Uncertain Data

Description Logic Programs (DLP) is an expressive but tractable subset of OWL. In this paper, we study a rising but under-researched problem of learning DLP from uncertain data. Current research rarely explores the plentiful uncertain data populating the Semantic Web. We handle uncertain data in Inductive Logic Programming (ILP) framework by modifying the performance evaluation criteria. We adopt a pseudo-log-likelihood based measure (PLLBM) which efficiently evaluates the performance of different literals under uncertainties. We implement our approach and evaluate it on two datasets. The experimental results demonstrate that our approach is able to automatically learn a rule-set from uncertain data with acceptable