Automatic Weight Generation and Class Predicate Stability in RDF Summary Graphs.
Mehmet Aydar, Serkan Ayvaz, Austin C. Melton · 2015
In this current study, we use graph localities and neighborhood similarity to enhance the summary graph generation approach for building a summary graph structure for intelligent exploration of semantic data. The key improvements to what we have previously proposed include the addition of a string similarity measure for the literal neighbors, development of a stability measure to evaluate the accuracy of class relations, the addition of auto-generated property weights, and the detection of noise properties.