As semantic web based applications are gaining popularity, very large RDF documents are becoming common. SPARQL is the de-facto standard in querying RDF data and research on efficient implementations of SPARQL interfaces for very large RDF graphs has attracted a great deal of interest in the recent years. However, in large datasets, the user faces the problem that the result set for her queries can be large. In this situation there is no clear for the user, from where to start looking at the results, since all of them are equally valid. Moreover, given the result of a SPARQL query, the only possible order is lexicographical which doesn’t help the user to distinguish which of the returned values should she look first. In this sense, it would be desirable to have a notion of “relevance” of nodes. A related problem is that of analyzing social network data. Most social network analysis concentrates heavily on finding social groups and finding the importance of individuals in a social network. However, this work generally considers the social network as a graph with a single type of connection, edges representing the existence of social communication or friendship for example. There are not many methods developed for social networks with many different types of semantic connections. As a result, there is very little work on querying of semantically rich social network data.