Diigo, API, RDF, semantic web - 114 views
started by Graham Perrin on 25 May 09
started by Graham Perrin on 25 May 09
started by Graham Perrin on 25 May 09
started by Graham Perrin on 25 May 09
http://groups.diigo.com/Web2/bookmark/tag/semantic
http://groups.diigo.com/Web2/bookmark/tag/RDF
Virtuoso Sponger (highlights) is particularly interesting.
I'm aware of (but not familiar with) the Dublin Core Metadata Initiative (DCMI). Some highlights from a quick search of their site:
http://dublincore.org/documents/dc-rdf/index.shtml
http://dublincore.org/documents/interoperability-levels/index.shtml
http://dublincore.org/documents/usageguide/glossary.shtml
Wish
I'd like a future version of Diigo to offer more to the semantic web.
Maybe, I'd like a more structured approach within Diigo to complement (not replace) the current free-form approaches.
This wish is very nebulous, sorry!
In terms of end results, what do I have in mind?
Well, for starters: Wolfram|Alpha sparked my imagination.
http://www.wolframalpha.com/gallery.html
http://www.wolframalpha.com/examples/
http://www.wolframalpha.com/examples/WebAndComputerSystems.html
At the moment I use it for mundane things such as
http://www.wolframalpha.com/input/?i=Brighton+weather+2009 but the possibilities of systems/services such as these are far greater. The overview video is well worth watching, if you're into that sort of thing.
Summary:
1. already, I love what Diigo allows me to do
2. I'd like a future Diigo, or Diigo + complementary system/service, to bring more structure/meaning to my content in Diigo, to groups' content in Diigo.
Brain storming (or just possible cerebral flatulence)
Diigo is perfectly situated to bring semantic tools to the public. Please convert the average Joe into an RDF contributor. Make it simple and, at first, stupid. E.g., a predefined vocabulary for common relationships: Manchester is a City; This particular Manchester refers to the one in NH, USA. Automate the rote work of translating to RDF.
Give the contributor the option to make meta data private, public but attributed (like a shared tag), or publicly available while the contributor remains anonymous to most users (not to admins to control spam, graffiti, and politically motivated flotsam).
For this to take off, Diigo must provide immediate benefit for Joe's effort; let Joe know that a topic of interest is being researched in Manchester, NH. Slap SPARKLE "magic" to Diigo's current objects (tags, comments, bookmarks, highlighting). Possible triple store entry: Joe uses tag XYZ. Use his existing contributions to feed the immediate payoff.
Eventually -- when people start to "get it" and want more flexibility -- allow them to customize their semantic vocabulary. Allow meta data and related vocabularies to be created collaboratively. Help out communities by plugging into tools like http://www.opencalais.com and existing ontologies. Open Calais makes inferences (guesses); let the user decide whether the guesses are correct and relevant. If so, a single click could add the Calais guess to the store of trusted knowledge. Anonymous feedback could be provided to engines like Calais to help them improve. Again, make the value obvious. Visually connect the Calais data with the important stuff on the page. Go crazy, even prioritize the connections by the user's current goal (what is the user currently interested in).
The highlighting feature already generates RDF like data. Example triple store entry: Joe cares about passage XYX on page PDQ. Add it to the triple store as such. The advantage of RDF over the existing highlighting feature will be for future searches. I'd like relevant portions of site to be "highlighted" before Joe sets eyes on it for the first time. With usage (and incremental refinement on Diigo's part) Joe's interest of the moment will be honored, the search will be short, and the browser's "find in page" feature will be obsolete.
And maybe someone on the Diigo team will be knighted, like Sir Tim.