Last Post

Farewell semantics... hello knowledge

I'm moving from this URL (semanticflux) to knowledgeflux.blogspot.com

Please follow me there :-)

Current PhD To-Do - Notes from supervisor meeting

So, on top of my day job... I have to do this :-)

Notes:

Reviewed what measures could possibly made directly from event log (i.e., an internal view). Need to consider how to bring in more “social” aspects. Need to have a concrete idea on what measures can be made on squidz data, and compare that to other systems.

Plan:

By next meeting, produce a report (format not important) which contains:

• Review papers on tagging papers, to identify (a) what features have been measured in other systems, and (b) what experimental methods can be cannibalised for squidz
• Details of what can be measured directly from within the Squidz log (i.e., precision, recall, click through rate, tags per annotation, tags per person, etc)
• Details of APIs for comparable tagging systems (Flickr, Delicious, Facebook).
• Details of what can be measured directly from the other systems
• Methods for comparing directly Squidz to other systems (with tentative ideas on how to explain the differences [squidz users have less tags, as each tag is more semantically richer])

Actions:

1. Nick. By next meeting. Produce draft report, as detailed above.
2. Nick. As soon as possible. Determine whether the report could be used as a basis for an ACTIVE deliverable on tagging systems (in order to justify spending more real time on PhD).
3. Nick. As soon as possible. To re-enable Essex username/password, to allow demonstration of squidz at next meeting

Issues:

Long term. Is it possible to migrate squidz onto an external server, for wider experimentation?

Date of next meeting:

About a month

Intranets vs. Internet

A perennial question is how to get social software running successfully for an enterprise?

Turning that on it's head, is the answer is why does it work on the Internet?

Personally, I would say that "size matters but failure doesn't". The mass of people available will mean that there will always be interesting content being generated, and people make new and innovative contributions.

Coupled with that is "why doesn't the intranet search engine return anything useful?" Again, size matters. Google, bing, AskJeeves all index billions of webpages; an enterprise would have a few million?

IBM seem to be going in the right direction. Have an internal sharing system (dogear), but is there a way of piggybacking of external sources of tags?

Hmm.

Now that is interesting

To do

So the current to do stands at:

• ACTIVate powerpoint and communicator - generate events from them, and send to context mapping process
• Work with Sandra and Marc on web service composition
• Still get Squidz in a state to launch -oh dear, I have to report to Dr B this afternoon of lack of progress
• Collate and write something about tagging - see “everything is misc”, “nudge”, "here comes everybody” and recent research papers on the meaning of tags… There is something here, but all I can see is a bundle of string, cotton wool and odd plastic soldiers. Motivation hits usage hits motivation hits community hits meaning hits feedback hits reputation etc.

Recent achievements:

• Submitted paper to BTTJ on “Semantic MediaWiki” - will be published in issue after next (June / July ?)
• Filed patent on automatic root cause and next event prediction in complex networks
• Reviewed ACTIVE deliverables
• Reviewed posters and demo submissions for European Semantic Web Conference
• Presented “Web2/Futures/ACTIVE project” to School of Business, at Kingston University. Now invited to be a speaker on the British Computer Society’s Young Professional programme.

Back again and beyond

So here we are again. Semantic soup reborn, and everything flux