Look Who’s Talking

Running the code in Example 3-14 creates a database that provides rapid access to message ID values grouped as discussion threads. With a database already containing the original messages themselves and another Lucene-backed database containing keyword search capabilities, you can really begin to do some very interesting things. Let’s now revisit Raptor from the previous section by considering the task of computing the individual sets of email addresses associated by any discussion thread where Raptor was mentioned, or, to put it another way, the various sets of people involved in discussions using this term. The resulting data structure we want to get to would look something like what’s shown in Example 3-17.