Stap isi

9 December 2008

Local information awareness

Or Bodging up a D.I.Y. DARPA for your LGA.

It’s a good idea to monitor mentions of your council and local area.

• You can learn a lot just by listening.
• You can respond to questions or issues in a timely fashion. (There’s also a growing expectation from people online that their comments about services or products will be picked up. Some companies respond to mentions with unsolicited customer support, on Twitter for example.)
• Internally you can promote the use of the net for community engagement and service delivery by raising awareness of the conversations happening there daily.

Here is one approach to automating the task, using Yahoo! Pipes to aggregate, filter and sort information from multiple sites and sources.

Eye on Dungog

I’ve picked Dungog Shire Council to use as an example, on the basis that I’ve had a beer at the Doug Walters bar at the Dungog RSL.

Pipes version 1 – simple

I ran a bunch of searches for Dungog, noted the RSS feed URL for the results of each, aggregated them in Pipes, sorted by date, and Doug’s your uncle.

Here’s the pipe and here’s the source.

Some notes:

• Specify a location when using Google advanced news search. Entering NSW in this field improved the relevance significantly.
• Flickr has a good search function but they don’t show an RSS feed on the results page. Use their public feed service and construct a URL according to those specs.
• There is a Flickr module in Pipes but it brings back the ‘most interesting’ results, rather than latest.

Pipes version 2 – smarter?

Definitely more complex. Using a larger set of keywords and phrases, a few more sources and a first try at filtering out common false positives.

Here’s the pipe and here’s the source.

Version 2 goes for a master list of keywords and phrases at the top of the pipe. The Loop module is used extensively, building up search queries for each keyword and phrase. Reason being, I can’t figure out how to chain keywords and phrases in Google search strings. And by keeping the input separate and distinct, I’m hoping you’ll clone the source, try it with keywords for your own LGA, and improve the pipe.

Some notes:

• For Flickr, this pipe just searches tags. But many people don’t tag extensively. You could use the flickr.photos.search API to include free text searches on title and description but this requires an API key (free).
• I’ve disconnected Icerocket blog search as it worked at first but then only returned errors. Google blog search also sometimes returns errors.
• Couldn’t work out how to get phrase searches on YouTube, so there’s a filter restricting input on that module to fairly unique keywords.

Results

You can subscribe to these pipes via RSS or have the results emailed to you.

— b3rn   listening, search   Dec 9, 09:57 PM   #

Comments

no HTML, use Textile

Name

E-mail

Website

Remember

• « Previous
• Next »