context in pieces

As it is one of my main intentions to (let the machine) draw evidence about a user’s social network and his individual sympathies from a rather dry collection of hard facts I will have a closer look on the data that can be accessed and collected.
The possiblity to log my communication behavior is based on the assumption that we communicate most of the time by technological means. Of course, this is not always true, which will require a chapter of its own: How do we communicate outside the “technological world” (talk, write letters, wave at someone, even give s.o. a wink), why do we do it, can it still be registered (by scanning the Bluetooth-Environment e.g.)?

In order to know what I am talking about, I want to take a closer look on the raw datafields available (as far as I recognized them), where I can get them from and whether they can be found in the datasets of the Helsinki Context Project already (cp)

datafields by medium

number of encounters phone call phone sms mail instant mes. physical cp
known person addressbook addressbook addressbook buddylist BT-log cp
social context sms: several recipients to, cc, bcc multiuser chat? other BTs around
direction initiator sender/recipient sender/recipient initiator
date/time phone call phone sms mail instant mes. physical cp
age of contact (known for how long, within the system?)
duration of encounter phone call length? filesize instant mes. (BT: difficult/slow) cp
place cellID, (GPS) cellID, (GPS) IP? IP? cellID, (GPS) cp
attachments? mail instant mes. exchange via BT?
profile loud, vibrate, … loud, vibrate, … status (BT active?) cp
calendar context (if available) if in use if in use calendar app? calendar app? on the mobile cp


There are some simple combinations that allow conclusions already:

  • frequency: number of encounters per time
  • regularity: a returning pattern over time (each Monday e.g.)
  • type of network: lots of contacts with few people vs. few contacts with lots of people
  • diversity: different media in use
  • type of node: passive (receiver) or active (sender)
  • groups: contacts often appearing together

Now, I will have to look out for a good parser (or write one on my own) and to get the ContextPhone client run on a Nokia N73.

[update 2007-05-28]

data fields by categories

The more interesting matrix maps the data fields to personal categories (that somehow reflect intimacy)

previewclip from the datamatrix

no Responses so far to context in pieces