Last night's topic for #pubmedia chat on twitter was station websites. Because I happen to have a list of public radio stations metadata (gathered from both NPR's station finder API and the PTFP Public Radio Coverage 2004 report, supplemented with FCC and Arbitron data), I thought it'd be interesting to quickly toss the results into a image gallery and see and compare the different sites. In about 40 minutes, I had a very basic, very ugly gallery up, and today I've relaunched it as a Ruby on Rails application still in development at http://stations.publicmediatech.com/. Because of the peculiarities of the data sources, there is some repetition of station websites (the data from PTFP is transmitter-based, so translators, state-wide networks, and other entities count as unique organizations) as well as a very broad definition of public media (including college, community and low power radio). The dataset is also missing a large swath of television-only broadcasters, although I hope to get a dataset shortly. My ideas for the future of the interface revolve mainly around making the sites more discoverable, including:
- Ingesting metadata into Solr, to support more powerful searching and faceting(implemented 9/4, using ruby-sunspot)
- Crawling the websites to extract full-text content (and some metadata) to support searching (e.g., last modified times, platforms and frameworks, etc) (first phase implemented 9/4)
- User generated content (tags, comments and ratings) to help organize the information. (implemented 8/31)
I took the screenshots using the nifty webkit2png python script (`for i in $( cat $f); do python webkit2png-0.5.py -D thumbs -d -s 0.5 $i; done;`) . The code to the interface is available through my github account, feel free to fork it and add features!