The three problems of content selection are: finding, filtering and bounding.
Finding has plenty of people working on it. There is Swoogle and (with Web3.0 annotation) even your regular WWW search engine could help.
Once you're found an initial starting point then filtering and bounding are needed. A SemWeb object could have many more triplets describing it then are needed, especially if indirect links are resolved and information is brought in from multiple ontologies. Filtering is the intra-object act of deciding which triplets are important. Bounding is the decisions in how far to traverse the graph of links between SemWeb objects. Bounding is the inter-object counterpart to filtering.