I’ve just attended a session called FaceTag: integrating bottom-up and top-down classification in a social tagging system. It was about research being done by Emanuele Quintarelli, Andrea Resmini, Luca Rosati.
I had planned to write up some notes on this session but rather than just re-tell Emanuele’s story I thought I would write a story of my own and incorporate some of his findings.
Here is a link to Emanuele’s slide show
Folksonomies or “cooperative classification” are terms used to describe how people can share ideas and information by annotating keywords to their pieces of information.
The problem I am trying to solve is not uncommon in professional services firm. It is the up hill battle when it comes to Knowledge Management. I’ll highlight some of the major challenges:
- Applying rich hierarchical classification is not natural to anyone who is not classically trained as a librarian (seriously I love librarians, where would we be without them?).
- It is incredibly hard work to fill out a lot of fields of information, often requiring the submitter to go back and re-analyze the item and even still, they may not know the answers to what they are being asked
- There is little or no incentive to contribute – this one may be coming around a bit, but you can be a complete non-contributor and be a workhorse (read: high revenue producer) and be rewarded handsomely, there is certainly no penalty for non-contribution.
Some of the ways around these challenges are by staffing for knoweldge management and in my organization, as I’m sure is the case in others, there are a few subject matter experts; former practitioners, who among other duties are tagged with managing the knowledge – they do this by:
- Culling through content to provide richer classification
- Harassing, nagging and otherwise dragging recent work product and information about it out of their colleagues
- Managing a hierarchical taxonomy for their area of competency
- Acting as the go-to person in their area of competency for the knowledge seekers they serve
Now that you understand the problem and some of the context around it I’ll take you on my own personal voyage, the one that ends when folksonomies answer some of the stuggles of KM.
I am responsible for the systems that support knowledge management, namely my firm’s intranet. I’ve watched the struggles of getting people to contribute over several years and while supporting these systems offered little in terms of solutions. Equally important I’ve watched the users struggle with using complex tools for categorization search and retreival, tools like document management sytems. My empathy towards users runs incredibly deep as many of these tools are hopelessly complex and entirely unusable.
I discovered folksonomies in early 2005 and spent a few months tinkering with them online for my own purposes and I began to realize that this may be a business solution and a user solution.
About a year after Del.icio.us, (an earlyish folksonomy for tagging Internet bookmarks) was launched, I tagged my first website with it. It was for http://Justcurio.us a since retired website, and my tags or rather tag was ‘folkonomies’. Shortly after learning and playing with del.icio.us I authored a short presentation on how folksonomies and cooperative classification could apply to personal information management and ultimately the greater knowledge management efforts.
I was pumped, I thought, right, I’ve cracked it, I know how to solve the KM conundrum of lack of incentive to contribute – it would be solved by individuals managing their own information with their own information retrieval needs in mind and then through serendipity enable access to others. Instead of billable time spent on search and retrieval, on email inquiries and distribution of documents, people would connect , collaborate and share ideas. It was perfect…..well, sort of.
There was one obvious problem. My audience does not produce web pages. They produced documents; deliverables for their clients in electronic or often printed format. And when identifying quality resources they open documents or print them to determine if they are applicable to their needs. So technically this needed some work – tagging from inside applications, uploading documents for tagging and various other scenarios, needed consideration. In addition to that barrier, I was completely consumed by another project and my visions of folksonomy grandeur slowly faded.
So, fast forward two years – here I am today, listening to Emanuele in his lovely Italian accent (I wish everyone who spoke to me had an Italian accent) talk about the deeper problems of information retrieval in a folksonomy. Problems of meaning. He sited the tag “food” in del.icio.us having many meanings to its taggers. He also discussed the impact, or lack of impact of a tag cloud on inferring meaning. He mentioned the issues of similar tags with the same meaning separating information and he also assailed the user experience of navigating a verticle list across thousands of pages.
The language challenge, where words have many meanings to many people, to me is an obvious problem on the public Internet. My favourite example is Ajax. It is : a way to web program, a household cleaner, a football team, a town in Ontario and probably other things I don’t even know about.
This is a major challenge to locating information, if you really want to know about the town of Ajax and can only find resources about computer programming or a Dutch football team you will be frustrated.
Emanuele asserts that the combination of traditional hierarchical classification and the user centered folksonomy approach to information categorization are the best possible solution to the deeper challenges of folksonomies. That tags plus information facets = accessible and retrievable information.
Now I assert in addition to this, that inside a company, where the domain of information is highly concentrated to one particular area, where there are employees who are focused on process and enablement and where the knowledge workers know each-other, folksonomies have the best chance of success and will contribute to business efficiency and knowledge management efforts greater than any other business initiatives targeting these areas.
This is because inside an organization, the risk of the “AJAX” problem is low. Everyone more or less shares the same needs in terms of information genres.
Also because employees can moderate and resolve the issues of duplication, synonyms and typos and add to the folksonomy in addition to providing richer and deeper hierarchical categorization that can be leveraged for many reasons.
And finally inside an organization you know who tagged something, you can call them, they are not a stranger to you – the connecting of people, the panacea of knowledge management, is finally on the horizon…
So my story is really not a story but a voyage that is ending with the following hypothesis. The hypothesis that folksonomies for personal knowledge management will be an incredibly useful way of creating rich, navigable webs of information – and if used will make KM, less about management and much more about shared knowledge.
If you find this interesting and are dealing with similar challenges, I’m interested in hearing from you.
