A while ago I digged around in my computer and collected _all_ the chat logs I was able to find: Trillian, MSN, Pidgin, Empathy, Thunderbird, XChat, Facebook, and normalized them to plain text. They were in all sorts of formats: line-by-line XML, JSON, per file JSON, per file XML, text, HTML. (I had to exclude Skype due to it's binary log.) If you have any logs, do a similar normalization as soon as possible, it will only get more painful. Anyway, the trillian logs were exceptionally ugly with with the line-by-line all data in attributes XML, but hey, the user interface and the connect to any platform was awesome.