TripleRater is one in a series of projects aiming to rate Linked Data triples by importance for human beings.
Linked Data currently includes a vast amount of facts (triples) about all kinds of things. Such structured data is great if you want to answer very specific / complex queries, as you'll only get very few and very relevant answers for your query.
But what if you actually only have a very unspecific query like "Give me some information about X"? In this case we actually have so many facts that we don't know which ones you might be interested in. We're trying to find out what to return in such cases.
Currently, we focus on approximating "importance for human beings" with human association strengths.
Please help us to collect some ground truth for human associations between semantic entities as described here.
If you have any questions, suggestions, want the data (before it's officially released) or want to collaborate... just contact me. I don't bite ;)