Ideas for sms-mining.

I was feeling bored, so I pulled all text-message data from my phone to my computer in a csv file. It contains date, time, whether it was out- or in-going, phone number and name of other person, and the text message itself. Since most separators, like commas, won't work because of confusion with the texts, I used pipes (|) (took a lot of regexes).

Now for the fun part. What could one learn from this? One thing I've thought of is the distribution of waitingtimes between messages. Now when my gf says I never answer, I have data to prove her wrong (or right). I guess it would be some independent poisson mixture, one for waitingtimes of days, and one for rapid back-and-forth messaging.

Another thing would be text-mining on the messages, like sentiment analysis and looking at trends, but that would take some work, and I'm looking for low-hanging fruit.

Has anyone done something like this before? I'd love to hear your ideas!

PS: I'm using R.