This is a quick heads up about something I was tinkering with over the weekend, and about which I'll be posting more in the next couple of weeks...
There has been quite a bit of talk, of late, in the blogosphere about "social search" - for example, Marissa Mayer in interview with Venturebeat or this interview with Jimmy Wales about Wikia.
One thing I've been trying to argue for for ages are custom, bounded search engines that search over sites linked to from OU courses. And over the weekend, the itch got so bad I had to scratch it...
So imagine if you will an OpenLearn unit such as T180_8 Living with the internet: keeping it safe. This unit explores the downside to "huge increases in ownership of home computers and ever-widening access [that has] been an obvious boon to many peoples' lives". In particular it contains many links to external, third party web sites, such as antivirus company Sophos, among others.
And also imagine a piece of magic code that extracts all the weblinks from this course and uses them to define the sites over which a Google custom search engine will search - in effect, a search engine defined by the sites referenced from this online course.
(You're familiar with the idea of a custom search engine, right? Like the How Do I? instructional video search engine, or the How Do Cook/What's Cooking? recipe search engine, or even this OpenLearn custom search engine...?)
Okay - so here's an example - a 'malware search engine' that searches only over the websites linked to from T180_8: T180_8 custom search engine.
PS the search engine will actually create a custom search engine on the fly for any OpenLearn unit using the URL http://ouseful.open.ac.uk/openlearnplayground/openlearncse.php?cc=unit_code
Is it possible to feed these custom search engines from deli.cio.us too? It would be really useful have a search engine that would search over certain tags and tag bundles and build these on the fly like your example for OpenLearn.
Posted by: Liam Green-Hughes at February 6, 2008 05:34 PMdelisearch was one of the first mashups i built (it was the driver behind searchfeedr.com, for example) and I've been keeping an eye on how that sort of approach has been evolving for ages.
For example, here's a Yahoo delisearch pipe (it will only search over about 32 domains, max, i think?): ./009301.html
here's an OPML generator that creates all the OPML you need for a delisearch pipe grazr interface:
./deli2opml2.html
Here's a variant I started to explore that used the tags as additional, automatic filter terms: ./010010.html
Here's another delicious grazr examples:
http://grazr.com/read/psychemedia/deliRolls.opml
This is what you really wanted though - a Google CSE generator for delicious feeds:
http://gadgets.kbac70.googlepages.com/csebuilder.html
(techie details here: http://kbac70.blogspot.com/2007/07/custom-search-engine-builder.html )