Hybrid Ranking and Context Extraction for Linked Data

Technical Report , Politecnico di Bari - 2010
The recent blow up of Linked Data freely available on the web calls for smarter methodologies and tools to query and explore datasets. In retrieval scenarios, there is the need for scalable techniques able to return also approximate results as a ranked set of promising alternatives. Among all available datasets in the Linked Open Data initiative, we focus on DBpedia and propose a new methodology to rank resources exploiting: (i) the graph-based nature of the underlying RDF structure, (ii) context independent semantic relations in the graph and (iii) external information sources such as classical search engine results and social tagging systems. To rank DBpedia resources, our approach refers to their context represented as a set of Wikipedia categories. We carry out experiments with real users to compare our approach with other RDF similarity measures. Based on this novel ranking methodology, we propose a semantic tagging system that recommends new tags semantically related to the ones initially proposed by the end-user.

Keywords: RDF ranking, Linked Data, DBpedia, Semantic Web, Semantic tagging, Content-based Recommendation

BibTex references


@TechReport{MRDD10a,
author = {Roberto Mirizzi and Azzurra Ragone and Tommaso {Di Noia} and Eugenio {Di Sciascio}},
title = "Hybrid Ranking and Context Extraction for Linked
Data",
institution = "Politecnico di Bari",
year = "2010",
url = "http://www-ictserv.poliba.it/publications/2010/MRD
D10a"
}

Other publications in the database

SisInf Lab - Information Systems Laboratory

Research group of Politecnico di Bari
Edoardo Orabona St, 4 Bari, Italy