This is pulled out of one of Ted's inline responses to the recent Setting up a recommender thread, and was hoping to confirm some things... Most of which may end up being a restatement of what he and others have said in the first place.
It seems that you would have a "document" in solr for each thing T that can be recommended, and each such thing would have a (solr) field for each type of item-item similarity you come up with ( llr from cooccurrence, the cross type stuff, whatever). Each such solr field consists of a list of the items "similar" to the particular item T for that field's notion of similarity. The "recent behavior as a query" means that we see which items the user has recently "interacted with" (clicked, viewed, or is somehow associated with), and search the solr field(s) with those items used in the solr query to find other items T to recommend.
