Abstract: In our participation in the TREC 2007 Question Answering (QA) track, we focused on three tasks. First, we processed the new blog corpus and converted it to formats which could be used by our QA system. Second, we rewrote the module interface code in Java in order to improve the maintainability of the system. And third, we added a new table stream which has learned associations between question properties and properties of candidate answers. In the three runs we submitted to the competition, we experimented with answer type checking and web re-ranking. In follow-up experiments we were able to further evaluate the contribution of these two factors, and to evaluate our new table lookup stream and combinations of streams.