2007-01-25

Second day of the Web n+1 workshop @ Sun Microsystems (Switzerland)

Andreas Blumauer of the Austrian semantic web school gave an overview presentation of semantic web technologies. I met him before at the KnowTech2006. He sees the rise of semantic web technologies as a result of the Glocalization: A mixture of Globalization (Expansion) and Localization (Specificity). In this world, knowledge workers integrate information in complex working environments, but this happens on a personal and local level. Information systems should provide an overview among this limited local level. He said that information systems should be able to answer the following questions, but are currently unable to do so: Who knows what, who knows who? Who knows somebody, who knows something about X?

I mentioned that the skillMap is actually capable of answering exactly these questions and he agreed. The rest of the day went along with overniew presentations of SW-related technologies such as RDF, OWL, SPARQL and D2RQ. I really learned a lot; D2RQ will possibly allow SPARQL-queries to the skillMap without the need to implement RDF om our side. Good news for a possible pilot with Sun.

Plane delay due to snow chaos postponed our arrival in berlin to midnight. I was totally exhusted when I arrived home. Poor Stephan's plane to Stuttgart was cancelled and he had to take a bus from Zurich and was home at 3.

First day of the Wb n+1 workshop @ Sun Microsystems Switzerland

Insights from the internal Web n+1 workshop at Sun microsystems in Zürich on 23/24 january that we were kindly invited to by Peter Reiser.

On the first day, we received positive feedback from the Sun people on our skillMap/tagNet project that I demonstrated twice during that day. Some quotes of participants' reactions:

"I think its really cool"
"they're onto a real problem"
"their model is the better model"
"It would be useful to pilot"

That was a great start and we hope to get approval for a pilot that we planned during the afternoon. High spirits in the evening, great swiss Käsefondue at Swiss Chuchi in central Zürich. A big thank you to Peter for the kind invitation and a tremendous day. I met some great people, for example Matt Stevens, Henry Story, Dave Levy and Chris Gerhard. I got a short mention in Dave's blog.

Labels: , , , , , , ,

skillMap / tagNet project and tagNet vision

I checked with Stephan and he says its ok to post the following description on the project we're currently working on: We're developing a technology that enables a community to jointly connect tags with each other with reference to their relation. In a playful graphical interface that is based on preFuse, users can draw lines between tags indicating how they're connected. In this way, a semantic structure of tags is created over time in a community-based bottom-up approach. We call this the tagNet. tagNets allow semantic search, mtching algorithms and similarity ratings for obejcts that have networked tags associated to them.

Sarah and me developed the tagNet vision based on her original vision of a knowledge management system that we call skillMap. It is a java applet that implemented the concept of networked skills in an application we call the skillMap. In the skillMap, there is two networks: a social network of people and a taxonomy of their skills. People tag their own skils and everyone in the community networks these skill into a loose semantic structure. The resulting data structure is a connection of a social network and a bottom-up skills folksonomy. The application of a similarity algorithm I am working on in my PhD thesis on this structure can answer questions like Who knows what, who knows who? Who knows somebody, who knows something about X? Who should talk to whom? Where are communities of practice?

My PhD thesis

Time to sum up the process. Since updating my page at the institute is so cumbersome, I can as well do it in my blog. My PhD thesis is about knowledge: what it is, how it's represented and how to measure it. It is also about how knowledge influences performance and how knowledge overlap in groups affects group performance. I don't really have a catchy title yet but it will be something like 'structural knowledge assessment and knowledge-related determinants of performance in complex problem solving' or something similar. It basically consists of the following parts:

1. What is knowledge and how is it conceptualized in the field of knowledge management? I've written a paper with Kozo Sugiyama on this, it will appear in the February issue of the Journal of Knowledge Management. One of its core assumptions is that unconscious access to structural knowledge is part of individual implicit memory.

2. Based on that assumption, I developed a computer-based psychometric test, the association structure test, that derives a knowledge graph for a given stimulus based on free term associations and reaction times. Unfortunately, it doesn't work quite as good as I expected as I explain in a paper that I successfully submitted to the WM2007 conference. I am also co-chairing the workshop on new approaches for implicit knowledge in knowledge management (NAIK2007) with Wolfgang Scholl at WM2007.

3. Based on structural assumptions from (1), I have developed a similarity algorithm for the skillMap. The skillMap started as a Knowledge Management System envisioned by Dr. Sarah Spiekermann and I have contributed to its actual design and development. There is an own post on that project, but the algorithm calculates a similarity measure between persons based on their knowledge self-assessment. That seems to work pretty good, but I currently cannot publish anything on this as we're trying to spin off a company and sell the technology.

4. Prof. Scholl, the supervisor of my PhD, formulated some theories on how knowledge overlap inside teams influences their performance. Hy says that there is a second-order polynomial connection between cognitive similarities of team members, knowledge increase and performance. I was able to validate his claim with an experiment I conducted in summer 2006. The paper on this is completed and I am currently waiting on his feedback prior to submitting it.

I hope to turn it in in June...

Labels: , , , , ,