I’ve been reading a number of blogs including floatingsheep.org and Pete Warden’s blog. Both of them occassionally have very pretty charts, the likes of which I would like to make on my own to geographic map patterns for Australian sports culture. Some of the maps that geographers do to look at social patterns are just fascinating in the extreme.
One of these geography related maps looked at YouTube. This inspired me to look at YouTube. I quickly realized that for big, popular teams, looking at YouTube would not be very useful. I do all my data mining by hand. (This includes looking up things like Brizzy and Wellywood after getting the data tableized to be able to identify a person as living in Brisbane, Queensland and Wellington, New Zealand.) It is time consuming. For smaller teams though, it seemed like the sample was small enough that I could get some interesting data off YouTube. This only includes videos, not playlists. At a later time, this may change.
This is both correct and incorrect. It turns out that most people are not listing their hometown on YouTube. They will often list their country though.
Anyway, the methodology I decided to use for YouTube data gathering was as follows: Search the team name. Find videos that pertain directly to that team. Look at and record the profile information for all individuals who have a video which mention that team. This should give an idea as to league interest and team interest, though not always. You can get say a Geelong Cats fan who uploads a video where the Geelong Cats play the Hawthorn Hawks and because of how I’m searching, they’d be included for both teams. It can be make interests slightly misleading, though fans of both teams will probably watch a video labeled that way.
One problem I’ve discovered (and I’m starting with the least popular teams first) involves the Geelong Cats. I’ve mentioned them before as they are problematic in that their VFL team shares the same name. Unlike LiveJournal and bebo where interest search makes it hard/impossible to separate the two, it is a bit more doable on YouTube as all you have to do is add a VFL to the search. Except, well, the VFL was the precursor to the AFL and there are a fair number of videos for the VFL Geelong Cats in the days before the AFL existed. I can separate these out a bit… but I’m opting not to at the moment. If a video mentions the VFL Geelong Cats in the days before the AFL, I’ll count the person who uploaded it under both the AFL and VFL.
I feel like I should almost be recording information about specific videos on YouTube but that doesn’t feel practical. It might be worth noting the total number of videos like I’m noting the information about mailing list volume for Yahoo!Groups. I may eventually do that, but at the moment, it isn’t a priority. I’m more interested in demographic information of people uploading videos than in the history of these videos themselves.