Febuary 2007 - Full Swing
Rachelle's PhD Completion Seminar
My University has a 'public Completion Seminar' as part of the PhD experience (3 to 6 months before handing over the thesis for examination), where you can get grilled by your peers, supervisors and others. Surprising to many, a Phd is meant to be written in a language that is understandable by a lay-person in the field of speciality, so theses are sometimes highly readable. However, these seminars get more indepth, depending on the questions and who's putting them, giving the candidate an opportunity to refine their argument and the presentation of it to others. Rachelle's research is on Knowledgement Management, titled: "Towards a model to explain knowledge sharing in workgroups in complex settings". Mmm, this is prototypically the opposite take on knowledge as compared to the DigitalFriend emphasis on the Individual's priority, up front - so I went along with much interest... note these blog entries on various seminars, are far from comprehensive on what was said there, but merely a slight reflection with my take on a few of the issues put forward.
Her work references the 'SECI Model of Knowledge Creation' by Nonaka (2003). Li and Kettinger's 'Interative Knowledge Creation Model'. Draws upon Structuration Theory - structure and agency. The 'Theory of Distributed Cognition' to explain human cognition. She talked of knowledge embedded in artifacts and embodied in people. Other authors referenced were: Brown and Duguid, Lave, Wenger, Nahapiet, Thomas and Walsham. She highlighted two things essential for sharing knowledge: 'formal and informal social networks'; a 'shared information and knowledge based artifact network'.
Her own resultant model of Knowledge Management had Determinants and Moderators, amongst other things. Determinants were: CoPs (? - Communities of Practice?); knowledge advocates and brokers; incentives and rewards. Moderators were: facilitating mechanisms for social networks (FM-SN); facilitating mechanisms for artifact networks (FM-AN).
Examples of FM-SN's: SMS, MSN, eMail, groupware, wiki's, intranets...
Examples of FM-AN's: metadata, agent technology, web crawlers, ...
She mentioned that linking people to artifact networks facilitates "a strong sense of codifying what they know."
... That raises my problem again with traditional Knowledge Management from the DigitalFriend's perspective: With the DigitalFriend, we are focused on the individual and their immediate loved ones, while their employment and related responsibilities come second place or further down the priority list in a person's life. Traditional Knowledge Management on the other hand, is so often about extracting it from employees heads and codifying it within corporate systems. Business books aimed at CEO's and upper management (usually written by one) put the options forward: a company might be "run for shareholders or stakeholders". Whether the book reader is considering downsizing, offshoring, rightsizing, outsourcing or just business process re-engineering, it will invariably ask, usually in less blatant language than what follows, but with the same intent: should one develop 'trust' in order to extract the knowledge from employee heads? i.e. 'Trust' has so often become a corporate tool, rather than being real trust. Therefore, if you start codifying your knowledge at all, it needs to be in 'your personal' technology, owned by you, controlled by you - preferably that you can carry around in your pocket. i.e. It should be in your DigitalFriend rather than in corporate Office tools.
I'm teaching a subject called Information Modelling & Database Systems again this semester, and today was the cut-off day for getting the class notes to the printers for subsequent publication and printing. Its a Masters level subject we give at the University of Melbourne which I've taught now for 8 semesters straight, and perhaps surprisingly, my enthusiasm in doing it continues to climb. I've decided to compile my copious notes, modelling example problems and their numerous answers into a text book on the subject, then track down a publisher. The idea is to capitalise on the time and enthusiasm, to use the energy during the giving of the course, to output a book.
There's 23 lectures (and assorted labs, tutorials and assignments) and I decided to make the focus of each lecture a book chapter. I drafted an introduction to each chapter, beginning with only a page or two on my primary motivations for each chapter. They are currently rough and brief, and in an unedited brain-dump format. I have very strong views on information modelling - it is also a significant aspect of my current PhD studies within agent oriented systems - so I've put up that rough brain-dump from chapter 4 here, to see if it strikes a chord with anyone:
Chapter 4: Information Modelling
'Once it is conceivable, it is achievable'. Anon .
Modelling is currently seen as a specialist activity, a niche occupation suited to certain top-down conceptual thinkers. It will become something that everyone can and will do at some level. I liken it to the map-makers and navigators in the Age of Exploration - these people were specialists, this was a niche occupation for the select few. Now, in our cars, trains and planes, the generic modern person is an explorer and we all need some of the skills of the navigator, the route planner, and reading maps and street directories is an everyday, anybody sort of activity. The same will happen with conceptual information modelling.
When I do a good data model, I often get this feeling that many long-term technologists have from time-to-time. I get a shiver down my spine as the thought crosses my mind that 'this data modelling process has a big and ongoing future'. I've learned to trust that shiver as a vindication of the thought that preceded it. For example, in the early days of sparse PC ownership, most of us using them knew that eventually everybody everywhere would be making use of these things in most aspects of their lives. I recall going to a local PC User Group meeting in Melbourne in the late 1980s which was addressed by Bill Gates. It was in a ground-level theatre in the Rialto building complex in the centre of the city, with exit doors that spill out onto the street. I recall Bill Gates at the conclusion of his talk just walking through the group of members, talking to one here and there on his way out onto the street and off down through the crowd of office workers he went, heading home. His personal worth was only about $100m then, and virtually nobody outside of the computer industry had heard of him, and far fewer knew what he looked like. I remember thinking "I hope he's enjoying his virtual anonymity, as soon the whole world will have a PC of some sort, everyone will know who he is and how much he's worth."
Why the divergent Bill Gates anecdote! Because the way I feel about Information Modelling, is similar to the way I felt then about the PC then - that everybody will be doing it at some level in the future. The data sources people will draw from are being delivered by the Internet's content providers via the open standards of the www. Anybody and everybody is putting information up, for others to consume and use at will. Note: Time Magazine's Person of the Year in 2006 was You - as the content providers and bloggers of the www in the current guise of Web 2.0.
The semantic web mentality - which presupposes that information modelling can all be automated - overlooks the ingenuity and innovativeness of humans to keep on coming up with new ways to align sub-sets of information from that which is commonly available, and generate new knowledge in the process. As more information in scope, richness, complexity and cultural diversity makes its way to being freely available on the web, the information modeller will be just one facit of the ever complex new renaissance wo/man. New sources and more variety of information will be put out there, and creative new data models and applications which use them will proliferate in ever-expanding and divergent combinations E.g. See the recent mash-up web applications as an early precursor to this trend. The web content provided by bloggers at this junction is really a parallel with the 'brochure-ware' of the early Web 1.0. In the near future everybody will be doing mash-ups at some level, and the good ones will be based on well-modelled information structures.
RSS Feeds in the Knowledge Tree
I'm storing the actual RSS channel information and RSS item information in a central file, so that they can be represented in and accessed from multiple directories in the Knowledge Tree. This requires both of them to have a unique identifier. Channel is easy, I just give it a Channel ID, and then need to be aware if the user attempts to include the same channel as a new channel. i.e. I generate the Channel ID from the feeds url.
Item is a bit harder, there are any number of them per Channel, and there is no formal identifier in the standard. The <title> and <pubDate> together appear to offer the best chance of a unique identifier, so I use the two of them concatenated (if both available), to generate a unique key. Keeping in mind the RSS standard says that "All elements (of Item) are optional, however at least one of title or description must be present." In addition, <description> may contain the whole news story, rather than having a <link> to it. Therefore, if <title> is absent, I use the first 255 characters of <description> to generate the identifying key.
Photo 1: The Role Message Lens.
The Role message lens is the third use of the FUN interface within the DigitalFriend (after the 'agent dial' and the Knowledge Tree). It displays all 76 generic roles and sub-roles in the Shadowboard methodology. It is used to filter/highlight just those messages in the user message window, which were sent to the user from agents that are assigned to that role or sub-roles in the user's life.
I better chase up the new icons for all 76 roles, as I've not troubled the graphic designer while the bug was in the interface.
Now for the finishing touches to saving RSS feeds in the Knowledge tree...