mySociety

Home

How should we handle categorisation for petitions?

Written by on November 20th, 2006 in Developers, No10 Petitions Website

So, there are now over 600 petitions in the petitions system, and we’re getting a steady stream of appeals from our users to add categories.

I’m posting to ask how you all think we should handle this. It seems to me that there are a few options:

  • Ask petition creators to pick one very basic top level category of no more than 10 or so, taken from a hierarchical taxonomy like the one the BBC uses.
  • Ask petition creators to pick the top level and the subsequent sub-levels to be more specific.
  • Go all web 2.0 and simply ask people to tag their petitions with some key words

More than just thinking about the overall philsophy I’d also appreciate thoughts on design. When you come to the homepage, how should the category system be presented to you? Tricky stuff, and I’d really appreciate your thoughts.

6 Comments

  1. Phil Rodgers says:

    During the petition creation process, do some sort of keyword matching to list other petitions that appear to be on the same subject, and ask the user if they want to sign one of those instead. This might eliminate some of the near-duplicates that are appearing.

    For categorisation, go for tagging, but see if you can get the No. 10 staff to do it rather than the users.

  2. Martin Belam says:

    Tagging by the users might be one way forward – but I can see two issues with it.

    One, I’d think it might be best for the petitions to *additonally* be tagged by the No 10 staff, giving a bit more control to the vocabulary

    Two, you might have a problem where a perfectly acceptable petition has to be rejected because of inappropriate use of tags e.g. ‘lyingbastard’

    Tagging also seems to work best on outlying nerdy sites that are seen as very clever (Flickr, del.icio.us), but that are not necessarily #1 for their niche amongst the mainstream users you are trying to entice into the democratic process here.

  3. Tom Loosemore says:

    yep, use term extractor then let users choose from tick boxes –

    better still, run different two ‘dumb’ term extractors across the body text, then only show those terms which both agree on, using tick box principle above

    better better still, do a term extractor which includes, but is not bounded by, a controlled vocab which No10 can update

  4. Renata Edge says:

    Hi,
    can I just get a word in edgeways as a non-techie? Do you have a proof-reading system? At the top of the page I spotted “heirarchical” (should be “hierarchical”) and a straightforward typo: philsophy.
    thanks
    Renata Edge

  5. Tom Steinberg says:

    OK, we’re actually going to start even simpler than any of these – just using the top tier of the ESD category tree (12 top level category names).

    Matthew’s going to add these as a simple pulldown menu, and we’ll see what the users say. We’re going 80:20 solution for the moment, so we’ll definitely come back to these more comprehensive ideas if the simple one’s no good.

    Also, thanks for the typo tip. I’ll try to fix.

    Tom

  6. Tom Longley says:

    For inspiration: Ask MeFi has a “hybrid” admin-defined and user-defined system for categorising its content. The user has to choose one of around 10 pre-defined categories, and can then add their own arbitrary term to further refine the categorisation. I think this make the initial navigation of content far more stable and commonsense than a full-on folksonomy.

RSS feeds

Categories

Recent Posts