I have noticed that besides the usual data publishing cases like a gov/ngo website with csv exports there are also services like
http://www.infochimps.com/
http://www.junar.com/
http://knoema.com/
where some of the data is aggregated and is made easier to interact with - especially if you need to combine multiple sources which is often the case. In a similar ways we can say that a standardized format of the data itself greatly improves the accessibility and reusability.<p>These services often come with bundled ways of visualizing/combining the data. These seem, unlike the data, siloed in each particular service/software. I feel that there should be a standardized form/protokol (eg HTTP) through which algorithms could be made reusable in the same way as the data is.<p>Basically I mean something like Weka+Vis as a Service hosted by a central quality assuring entity provided together with a data marketplace based on a common algorithm form. It would serve as a marketplace for algo/mining/vis. designers, as a testbed for researchers of new approaches. Customers would just put their data in and picked what processing they want, or they could call the algos through a common HTTP API with data on their side. Onsite dedicated solution for enterprise. I realize that there are many issues with a common algorithm form because of the intrinsic variation, but maybe differen interfaces for different classes of algorithms (eg stream based, batch...) could help.<p>There are too many great things getting published in academia in this space that never see any real world usage (or re-use) and there is no place I know of where I could try out my ideas meaningfully.. I could implement stuff, put in some data and publish on my blog - but how is that actually useful for people who need to make sense of their data?<p>Thanks for any comments or pointers to existing projects!
~ an aspiring datavis practitioner