Research talk:Editor lifecycle

From Meta, a Wikimedia project coordination wiki

Hi, are you planning to just do this in terms of editing activity per day/week or are you going to look at things like types of activities and which namespace they get involved in? If so Soxred's tool might be worth looking at. See an example, though I think that particular tool is opt in only as the raw results get posted onwiki. WereSpielChequers 11:47, 7 July 2011 (UTC)Reply[reply]


We have some longstanding editors who already had bot in their name before the bot rule came in. We also have bots that don't contain the word bot. I suggest that as well as flags you use this list of unregistered bots en:Wikipedia:List of Wikipedians by number of edits/Unflagged bots. WereSpielChequers 21:23, 13 July 2011 (UTC)Reply[reply]

Thanks! I just found one bot (Guanabot2) in the 2004 cohort that escaped the filter. As for the legitimate users with 'bot' in their user name the pattern matches the string at either word's ends (e.g. Botblah or Mybot would match, but Abbott would not). Still, there will always be a certain rate of false positives even in this way, but it's not going to affect the overall statistics as much as a bot would do. So it is better to have a classification scheme that errs on the side of overzealousness than the other way round (i.e. does not filter out any legitimate user, but have some bots slip thru). Junkie.dolphin 17:36, 14 July 2011 (UTC)Reply[reply]