[okfn-labs] Ideas for data mashups and improved visualisation for CKAN

Gavin Chait gavin.chait at okfn.org
Wed Jul 24 16:08:33 UTC 2013

Hi Friedrich,

Thanks for the feedback and I appreciate exactly that many people may not want to embed directly from a government site.  The additional components you mention, JS pivot tables and a mashup recommender, sound brilliant  too.

I don’t think the intention is to glue everything together, but the current approach of download data from one place, do the analysis/mashup in a spreadsheet, then upload to a variety of places to achieve different visual ends does create a lot of work.

CKAN has an API that should facilitate integration with these different platforms, whether they are deployed with CKAN or the data is piped through to them during search.



Gavin Chait
Head of Services | skype: whythawk | M:  +44 (0) 78 9495 7090 | @GavinChait
The Open Knowledge Foundation
Empowering through Open Knowledge
http://okfn.org/ | @okfn | OKF on Facebook | Blog | Newsletter

From: friedrich.lindenberg at gmail.com [mailto:friedrich.lindenberg at gmail.com] On Behalf Of Friedrich Lindenberg
Sent: 24 July 2013 15:35
To: Gavin Chait
Cc: okfn-labs; CKAN discuss
Subject: Re: [okfn-labs] Ideas for data mashups and improved visualisation for CKAN

Hey Gavin, 

I think its great you're soliciting input, but would also consider asking the ddj and school of data communities, as they are closer to the user end of the spectrum. It seems really important to base this off concrete use cases rather than the assumption that "people" will want to embed "visualizations" to "tell a story" (this doesn't mean anything). In CKANs specific case, there's further fun in the fact that I don't know how many people would actually be happy to directly embed a visualization from a government site - it seems unlikely to me, for example, that many journalists would do that. 

I would also note that the best data vis tools I know focus on doing one thing and doing it well. Examples include DataWrapper, CartoDB, Sigma.js, Timeline.js and most of the other amazing stuff on http://selection.datavisualization.ch/. The "we'll glue everything together so that normal people can use it" thing hasn't worked for anyone (Socrata, Fusion Tables, ... but least of all: for the users). I know this is a great temptation, but it's really just an escape into meta-problem-land when there are plenty non-meta ones to solve. 

For CKAN, I think there are two tools that I'd love to see personally: 

1) A really nice JS pivot tables implementation. When you're trying to do some exploratory analysis on data, the best thing you can get is a table. Even better is a table of aggregates. What I would love to have is a small, stand-alone JS library that allows me to pivot, filter and sort some data. If this had some flexibility wrt its backend, it would be extra-awesome (OS needs this, too). And no, facets are not really the same thing and Recline doesn't do that now :) 

2) A mashup recommender. CKAN has the unique advantage of often holding many related datasets, and Martin and I have both done some research into whether it could generate recommendations on which of these datasets can be combined into a new mashup. I realize this isn't really vis, but it belongs into the pipeline and would be pretty useful to help the process. 


 - Friedrich 

On Wed, Jul 24, 2013 at 12:38 PM, Gavin Chait <gavin.chait at okfn.org> wrote:
Hi all,

We would like to massively improve the data explorer and visualisation
functionality on CKAN and are looking at open-source libraries and toolkits
we may be able to use.

Our brief is as follows:

Search and select data from CKAN, overlay datasets the data to create
mashups, and present these in an engaging and easy-to-use embeddable visual
format.  Data would also include geospatial. Data could include transport,
air quality, energy usage / resource efficiency, and licensing of premises /

There have been some cool new initiatives from infogr.am, Dataseed and
Datawrapper.de. There are also some great geospatial visualisation
libraries, like Kartograph.

Our interest is in open source solutions to look at:

1. Managing diverse data alignment for the mashup;
2. Outputting that data to an embeddable visualisation.

We can build on Recline and improve it, or we can work with another existing
library that offers us greater flexibility.  We don't want to start from
scratch, though.

We would love to hear your ideas and suggestions as we look to the next
generation of CKAN data explorer.

Thanks and appreciated


Gavin Chait
Head of Services | skype: whythawk | M:  +44 (0) 78 9495 7090 | @GavinChait
The Open Knowledge Foundation
Empowering through Open Knowledge

okfn-labs mailing list
okfn-labs at lists.okfn.org
Unsubscribe: http://lists.okfn.org/mailman/options/okfn-labs

More information about the okfn-labs mailing list