Principles of Distributed Representation (SAC2005)

Principles of Distributed Representation
Stephen Downes

“Not another metadata talk” he promises…. since Stephen puts his presentations online, and I saw him turn on Audacity to record his own session, I am bowing out of taking copious notes (well no, I changed my mind, see below)

While the intros are going, Stephen has a nice slide show of his photos from a visit to Aspen.

Before the notes, it’s worth saying that Stephen in person is how he is in hos written work – deeply thoughtful, self effacing, sarcastic, humorous, challenging…) also in his presentation, note the use of references and information just published yesterday.

And Stephen was on spot- challenging us, saying boldly what is crap and what is not… curious to hear from others how it was heard.

Onto a few notes (Cyprien Lomas was taking more detailed notes than I, but trying to do it together in SubEthaEdit, we ended up stomping on each other…)

Starting with a philosophical tact…

Making a point about knowledge not being just transaction, like sentences… “what we know depends on our point of view” — but that is all we have– language is at best an approximation of reality and at worst is a paradoy of knowledge.

Knowledge is sub-symbolic — mere possession of the words is not the same sa knowledge, makes sense when applied in context– a lot of knowledge cannot be expressed in words (e.g. skill of how to throw a dart).

Knowledge is distributed- no person that possesses this knowledge. This morning the space shuttle landed. Where does the knowledge of how to operate and land this vehicle reside? It cannot be in one place or person.

Knowledge is interconnected.

Knowledge is personal – internal associations… “How many of you thought of ‘plaster’ when I said “Paris”? Who thought of ‘Hilton’?

Knowledge is emergent- not contained but emerges as result of connections between minds. Emergence requires perception.

Wisdom of Crowds. Each of us has a piece of the puzzle- not acquired, we create it. Creation is process of acquisition, remixing, repurposing, feeding forward.

Knowledge is the interplay between our bit of the network and other bits of the network.

Comments that his talk actually evolved from ideas in the opening keynote yesterday from Charles Vest “You thought a presentation was static”

The idea that our different activities are distinct is fundamentally misguided (no separation between home, work, school)- interwoven. We have metadata that is “school” and others that are “school” and they will never meet!

Open – the idea that we can store knowledge in closed repositories is fundamentally misguided (‘requiring passwords, authentication, blood types’). If knowledge is connected like neurons in your mind, than putting up barriers like passwords, copyright is an “illness”.

What about metadata?
* not merely properties of universities
* basic unit of knowledge isn’t the university

Connection of entities, each acting autonomously, connected in an open network. Knowledge consists in the connections between nodes. Same interaction between bloggers is same as networks of rivers, the phenomena of crickets chirping (all networked).

“That’s the background how I approach meta data”

Resource Profiles-
* different types of metadata (microformats) [diversity]
* info is distributed [interwoven]
* any given perspective is aggregated [open]

Learning Object Metadata (LOM) a “standard” it is the oddest thing in the world for every object to have the same exact structure- one big monolithic format, and we end up mis-shaping the data.
* nature of LOM should vary according to resource
– technical metadata, use audio elements (bitrate) for sound
– use video elements (framerate) for video

e.g. objects should have different kinds of metadata depending on what it is, how used.

LOM needs to take its place among a wide range of other metadata types
* personal FOAF
* rights ODRL
* events, organizations, publications, more

Must stop seeing it as a separate entity- ideal metadata is RSS- title, description, link, simple, beautiful. Our idea of what a learning resource could be is limited by LOM- how does an event fit in? Where is the field for “start time”?

LOM collapses our view of what a learning resource can be to something limited.

How would you describe a person in metadata?

Distributed Resource Descriptions
* info not only in person c.v. but in school transcripts, police records, credit bureau, phone book, etc…
* a prospective employer (would you depend solely on a c.v.?) is interested in a very different profile than a prospective date (point of view)

What we know about a person is not contained in a single metadata record or a location. Information is aggregated, remixed to meet their perspective.

Metadata Referencing
* metadata about a given resource is not stored in a single file
* references to external entities (authors, licenses) are created by pointers (rather than stuffing it in each metadata); e.g links to creative commons, rather than embedding a creator of a learning object in metadata (what happens to 20,000 metadata files when a person changes their contact info) rather than point to a personal metadata
* in RSS, the “rel” data entity (embedded in web pages)

Principles of distributed metadata (== database normalization)
* metadata for a given entity should never be stored in more than one place (still allows mirrors, caching)
* metadata for a given entity should not contain metadata for a second entity

Not Just Metadata
* concept applies to learning resources.. anybody can produce resources, anyone can use them
* also applies to applications, a learning environment but many
* small peices loosely joined

Web is changing…
* Web 2.0 – not just a slogan (shift from web as medium to web as platform, or environment)
* shift from idea of web as communication (information theory) to the idea of it being a place.

E-Learning 2.0 (Scott Wilson, David Tosh diagram)
http://www.cetis.ac.uk/members/scott/entries/20050125170206

Future learning environment in middle connected to other types of external systems (outside a university system), e.g. flickr, 43 things….

Learning as a network phenomena

Microformats on the web (http://microformats.org/)

Web 2.0 checklist:
* structured microcontent
* data outside, not local (feeds galore)
* web APIs (e.g. flickr, Google Maps)
* single identity
* user generated, user managed

1st part metadata – created by author
2nd party metadata – created by user of resource (reviews, how used in context)
3rd party metadata – created by an observer

Closing:
* Charles Vest talked about meta University
* This (what I described today) is the info architecture for it
* The key is not large integrated systems, but rather small flexible bits that can be connected

LOM will be written, or bypassed entirely, instead of a single standard, will be a set of microformats, will link to other types of metadata describing authors, resources, rights, etc.

If this kind of stuff has value, please support me by tossing a one time PayPal kibble or monthly on Patreon