Page MenuHomePhabricator

Pywikibot-catimages: Meeting 1 - 1 May 2016 - 13:00 UTC
Closed, ResolvedPublic

Description

Date: 1 May 2016
Time: 13:00 UTC

Description: First meeting to discuss the project and talk about the plan and lay down norms of working.

Agenda:

  • Meet the first time
  • Consensus on plan and timeline (project plan)
  • Decide on weekly meeting timings - 6pm India
  • Suggestions on ways to do 'community bonding'
  • Create a common workflow (norms of working, eg: Review procedure of code)
  • Focus for python2 or python 3 ?
  • List of tags actually needed in commons (Review of list of things done by catimages)
  • Priority of file types (jpg, png, audio, video, pdf, etc)

Minutes of the Meeting:

  • Project plan details will be decided a little later as scope of catimages is not completely known yet. This can be a changing plan, but the rough plan in the proposal should be followed. Other than that, the weekly objectives decided in the weekly meetings should be accomplished.
  • Meetings:
    • Weekly meetings in IRC (channel # gsoc-catimages): Sat 12:30 UTC (18:00 India, 14:30 CEST)
    • Every second week in Skype: Sat 12:30 UTC (18:00 India, 14:30 CEST)
    • General (faster) communication by: IRC, mail
  • Community bonding objectives:
    • Need to get in touch with the pywikibot and commons community.
    • The pywikibot community has been done to some extent. Solving more bugs and doing more code reviews would be great.
    • Bonding with commons needs to be worked on. Need to do edits, file uploads, etc on commons. Attempt to target 20-100 useful edits per day for this week.
    • Possible tech related tasks in commons:
      • Handle categories related to Unidentified people
      • The Wikipedia Zero program in Bangladesh and Angola being misused using fake formats: T129845 and T131934
      • Identify whether a photo is meant to be in landscape or portrait using OpenCV
  • There is no specific common workflow decided yet. Public Github repos will be used for now where no specific code review and so on will be carried out. Later when merging it or using it, this will be decided.
  • Python2 was considered the main target. Mainly because pywikibot-core is not going to stop supporting python 2.7 any time soon. Hence, Python3 should be kept in mind when coding, but python2 is the main target for this GSoC atleast.
  • There is a list of all metadata and categories which are analyzed by catimages here - https://etherpad.wikimedia.org/p/Zl7V7KuK7J
    • Need to get in touch with the commons community to understand better what categories they like.
    • Understand what sort of categories are needed by commons.

Event Timeline

DrTrigon updated the task description. (Show Details)

I created {E172} for this meeting (but the calendar app is not very functional yet)

I've sent out a google calendar meeting request to everyone (and then updated it because I had the timezone adjustment wrong.) Hopefully we're on the same page now, across the timezones.

Notes to me:

  • project "charter", plan, etc. (in what detail?)
  • meetings, e.g. Fri "afternoon" (UTC >= 12) and on need (e.g. through IRC) - may be have an "alarm" in case of getting totally stuck
  • have no experience with py3 yet - but think this is essential here - whats the state of pywikibot?
  • priorities; good question - pick first the low hanging fruits (follow the 80-20 rule; https://en.wikipedia.org/wiki/Pareto_principle)
  • to what extend is it useful to make old code running? local, wikilabs
  • start by separating bot and framework and make them running?
  • ...then go to py3? ...or do this as very first thing?
AbdealiJK updated the task description. (Show Details)

Intressting... Thanks jayvdb!
My guess: In order to be able to properly takle the problem, we would also have to surveil the FB pages for links posted and do a cross correlation analysis... Need to go to the source instead of trying to clog the sink.
I think what I want to say is; do NOT make this 1. priority. E.g. starting with the faces bussiness should be way simpler as it is already there. That way you can already get some edits and show you are able to control "the beast"... ;)

Possible concept could be to have a bot similar to the interwiki.py bot going through groups and users (build a tree/graph and go to all nodes) and noting any suspicious behaviour. But anyways we should analyze meta-data and try to use it for an analysis...

http://stackoverflow.com/questions/22868139/how-to-get-all-content-posted-by-a-facebook-group-using-graph-api

http://facebook-sdk.readthedocs.io/en/latest/api.html