We recently finished development of a custom social search implementation using DotNetNuke and Venexus Search Engine for the System Center Central site. When performing Web queries, a fulltext index query is first run against the local database for keyword relevancy. If less than 15 results are returned, a Bing API Web query is performed (configurable via module settings). The results returned from Bing are added to the list, and then all results are sorted using our unique algorithm for social interaction for Rank.
When users click on search results, comment, vote, or rate the search results, the overall rank of the item increases or decreases depending on the social interaction. This is keyword relevancy searches combined with community controlled search results ranking, aka "social search". If you are interested in your own custom implementation, please contact us.