How to combine communication channel histories?

I regularly contact the same people, over and over again using a variety of e-mail, instant messenger and SMS. Conversations move seamlessly between these different mediums depending on where I or the people I’m communicating with are at the time. To my brain, the digital location of the conversation isn’t relevant, just the fact that I’ve been in communication with a certain person on a certain topic; if I’m lucky I might even remember vaguely when. So how do I go about finding where I had this conversation?

Desktop search tools are becoming better and better and able to index more and more data sources, so it’s likely I could search both my IM and e-mail logs from a single interface, but that still leaves my mobile device out in the cold, despite all its high-falutin’ Bluetooth and 3G/GPRS connectivity and l33t synchronisation skills.

Text messages on Nokia phones are stored in text files with some bytes at the beginning and end for denoting various details, as well as the number of the sender, for which a lookup to the contacts database is necessary to retrieve a useful name. But how to get this to the PC and indexed?

Maybe the best idea would be some kind of automated sync in the background to your PC over Bluetooth which transfers your sent and received text messages, and puts them into an mbox datastore. Of course, you could just transfer them as text files or maybe you could use a desktop search tool plugin (like the ones listed on the Google Desktop Search plugin-ins page) to identify them as a different type of data, but do you want to? Really you just want to search all of your messages in the same way so that “from:bob” returns recent emails, IMs and text messages from Bob (although I guess you should also have specific keywords so that “email:Bob”, “sms:”Bob” and “im:Bob” all work for their specific context for when you know the communication already).

The ability to search IM and email both as one and separately should be doable quite easily I think; the background-transfer of SMS information possibly less so – I’m not sure how the authentication issues would work and whether you’d need to actually take action on one of the ends (which would be less than ideal), in which case a manual sync might be preferable.

If anyone’s doing any work in this sector, or related, I’d be more than willing to lend them a hand, whether it be coding or just testing, but I rather think I’m too busy at the moment to start another project myself quite yet 🙂

Published by

11 thoughts on “How to combine communication channel histories?”

  1. Phil, that sounds incredibly powerful, and if I could be any help as well, should something like this come up, I’d be glad to put time towards it.

  2. I’ve been working on some scripts to archive my SMS messages as chatlog style messages with [YYYY/mm/dd hh:mm] Name: message style lines. I got myself some hacked piece of software that exports all of my messages to an easy to parse textfile, I transfer that file to my computer and have python scripts to parse it into nice file-per-person logs. By the time Python for Series 60 is a little further along, I should be able to cut the hacked thing out of the loop, and maybe automate it a little bit more.

    I have also been working on scripts that consolidate my chatlogs from different IM-clients, and I will be writing a script that takes sessions from the same person from different files and puts them in order in a new file.

  3. That sounds excellent manuzhai! I don’t suppose you want to drop me a line either here or via the contact form or email with a few more details about exactly what you’re doing?

  4. Teensy point:- you need a mechanism in there to infer author identity across the various channels, since unless you’re a very organised person (much more so than me, anyway) with equally organised friends, those on your various contact lists are variably known as, say:

    “Bob Whoever < flibble@bob.com> ” on email,

    “PricklesTheHedgehog” on GAIM, “B_Whoever” on Skype,

    and “Work Bob: 071234567890” on your mobile.

    You know that, but there’s no obligation for you to ever have formalised it anywhere. And it might get fun if you know several Bobs, or if a few of your friends have surprising nicknames. So there’s maybe some fun initial join-the-dots setup to plan for in there somewhere:

    “im:PricklesTheHedgehog” == “sms:Work Bob” == actual identity and so forth.

    Or I might just be plain wrong. Happens all the time.

  5. Yes, this is a very good point, and you’re definitely not wrong 🙂

    Fortunately I’m very organised about naming my contacts in the same way on different platforms, and the join-the-dots work is something I enjoy trying to solve in my free time 🙂

    Effectively, what I have is a FOAF file which lists lots of people in foaf:knows elements and uses the foaf:nick for the different ways I refer to them on different systems (based on their nicknames rather than their actual IM or email addresses, which I assign them in each app). When a new person crops up in my IM client or email contact list who isn’t already in my FOAF address book (which is, I guess, effectively what it is) I get prompted to add them in (I do this manually).

    There’s certainly no obligation for anyone to have performed this kind of formalisation before they want to start using such a tool, but it quickly becomes an invaluable bit of time investment (although probably doesn’t scale if you’re really popular :).

  6. Nice solution. Though I’m not organised at all. I’ve been using the same SIM since 1999, when some phones had strict limits on string length – so the newer the contact is, the closer the phone’s idea of their name approaches reality.

    Only thing is sometimes on stuff like IRC, the nick is sometimes appropriated for humour/drama eg.:

    /nick OsamaBinLaden
    <OsamaBinLaden> Stupidity is a religious duty, and we hope to be rewarded for it by God!

    Not that there’s not enough data to solve that, just a touch more effort than a plain grep.

    Leaving you only with the problem of how to extract sms messages from a phone without some horrifyingly complex manufacturer-specific solution (eg Oxygen).

  7. Crikey, a six-year old SIM. I’m not sure my annual-upgrade mind can understand that 🙂

    The IRC problem is an excellent point, but fortunately for me, one I don’t suffer from because I almost never use it 🙂

    So yes, we’re left again with getting the data off of the phone. I think it should be reasonably straightforward for Symbian phones, and possibly for modern SonyEricssons too. Anything else and you’re probably screwed. I mean, er, “locked in”. 🙂

  8. The Mac OS X address book can tie together phone numbers, email addresses and AIM login names, that’s the guts of em’s point handled (albeit only in one place).

    I’m sure my phone’s address book has an email entry too, although I don’t think I’ve ever used email from the phone.

  9. Yes, from the OS X perspective, this is almost a completely different prospect because the address book already hooks together Mail and iChat. Added to which I’m guessing that there’s something in Salling Clicker that lets you sync your contacts very easily with your SonyEricsson phone.

    Lucky I’ve not got one then, eh? 😉

  10. Moin, I use gsmlib on a linux box to get my SMS out of two mobile devices, via bloootoooth and serial/usb conection. Doing so I fetched about a megabyte in the last years, this 1MB text file is obviously indexable…

Comments are closed.