Archive for the ‘Uncategorized’ Category

Social Recommendations within the Multimedia Sharing Systems.

Monday, September 29th, 2008

Presentation by Katarzyna Musial at the 1st World Summit of the Open Knowledge Society, Athens, 24-26 September 2008. Track: Social & Humanistic Computing for the Knowledge Society: Emerging Technologies and Systems for the Society and Humanity.

Reliable Personalized Learning paths: The Contribution of Trust in E-Learning

Sunday, September 28th, 2008

Presentation by Vincenza Carchiolo at the 1st World Summit of the Open Knowledge Society, Athens, 24-26 September 2008. Track: Knowledge, Learning, Education, Learning Technologies, and eLearning for the Knowledge Society.

One of the major advantages of E-learning is the personalization of learning paths. But currently lifelong learning scenario often consists of a P2P network of prosumers (provider+consumers), so no central authority manages the learning objects. Thus the choice of the best learning materials as well as the best order (learning paths) is problematic.

The authors distinguish two types of trust: Trust among peers, which allows to express which one is more reliable (authoritative) in answering a query within a given topic (described by shared ontology). Trust peer-learn objects express which LOs are considered more useful by that peer.

Approach: The proposed model is based on a peer-to-peer network with prosumers and LOs (modelled as a directed multigraph). A peers stores his suggested learning paths LOS precedences-successions and relations) and assigns trust to peers and resources. LO is a resource described through its set of prerequisites and objectives. Peer-to-peer trust expresses the reliability a peer assigns to another peer about a given topic, the label is trustworthiness in the [-1;1] range. Peer-to-resource trust expresses the reliability a peer assigns to a resource. Learners can assign a query to search for reliable and personalized learning paths by providing: prerequisites and objectives; total time available; level of difficulty; and a threshold for trustworthiness. The results is a sequence of resources.

Related work: Ariadne, ALFAnet, Edutella.

This works is currently focusing on global learning scenarios without authority, not university scenarios. But it could be applied to an university-scenarios. Just that we would have to change the traditional way, in which students are consumers only. They should of course trust the lecturer, but should be encourage to contribute to the course, to enrich the course content with pointers to the web, new problems, and examples. Then trusting peers could help students to improve their lecture materials.

From “Publish or Parish” to “Demo or Parish”

Friday, September 26th, 2008

I discussed with other participants of the WSKS the situation in science regarding scientific publishing. Scientists have to prove high-impact publication (see science index), but publication processes are often very long. Consequently, results are often out-dated at the date of appearance.

A tendency that one may recognize is that more and more submissions focus on convincing that they are capable of doing something (realizing an idea) rather than displaying fully implemented and evaluated research systems. As I was told, this is also referred to a shift from “Publish or Die” towards “Demo or Die“.

If we look at the mass of scientific publications today, it gets harder and harder for scientists (in particular novice) to distinguish papers on fully evaluated algorithms and systems from vision-papers or work-in-progress as well as paper on still ongoing work from out-of-date publications. Would it be possible to classify papers more explicitly, i.e. to help (young) scientists to immediately know what to expect from a paper (ideally before having to purchase it)?

I agree that (to a certain degree) it is our task to learn how to make this distinction ourselves. For example, we can take the impact factors and reputation of the publication sources into account. However, although in these highly-recognized sources, the actual state of the presented work remains unclear (sometimes). This particularly refers to the presented systems, i.e. to decided whether a mock-up, a research prototype, or “almost-product” has been presented.

And overall, when is a good time to publish, i.e. what state should the research be in before it is “ready” for publication?

SCOOP workshop (Communities of Practice)

Thursday, August 30th, 2007

I am sitting the the SCOOP workshop of the JEM Network, which really shaping up nicely we have the MKM people meet with education and social software guys. I will blog a couple of impressions from the KWARC angle.

The discussion is quite stimulating.

Ralf Klamma (RWTH Aachen) gives an intro to Community Information Systems and claims that the constitutive features of CoPs are:

Mutual Engagement (ME): “You have to know which community you are belonging to”; I am a little sceptical whether this is really true for CoPs in Science which are very distributed, and may even be disconnected.

Joint Enterprize (JE): There is something you want to do together, and you want to learn to do it better. This is at the center at the CoP definition of Wenger. We have been neglecting this in our KWARC models here, or taking if for granted. We need to think more about this.

Joint Resources (JR): This is really where our MKM paper sits, and I have the feeling that we have something to bring to the table here. Klammer is interested in Multi-medial theories. I must say that with the OMDoc approach, we are interested in a Omni-Medial approach (OMDoc as a omnipresent semantic medium that covers all). The idea here is that the content Markup allows to generate multiple medial representations from this source and any media can be marked up to OMDoc. So maybe this is compatible.

Klamma also talks about a cross-media theory of transcription that sounds interesting (J”ager, Stanitzek Transkribieren – Media/Lecktu”ure 2002). The gist of it seems to be that events (e.g. historical) and objects are transcribed across media (e.g. to OMDoc or SciML). So we only have access to the media trace, not the event itself (it is long gone). I wonder what this theory predicts, it seems compatible with what we are doing.

A great example: The babylonian Thalmud has been transcribed to an XML markup, where you can annotate relations. Then the text can be acessed as semantic hypertext. One effect was that thalmud students were asking tougher questions earlier. That is very encouraging. I wonder if the sources are available for this, and how an OMDoc version of the thalmud would fare, and how much of the structure could be transferred in the CD-based structure we claim to be so essential.