Thursday, August 24, 2006

Last session of Day 1 of Hypertext

The last session for today is Novel Systems and Models. Jean-Yves Delort is presenting Identifying Commented Passages of Documents Using Implicit Hyperlinks. The technique that he is presenting is selecting passages from documents. Blog comments are used as implicit hyperlinks. Building prototypes of relevant comments is difficult because there are many different types of comments and the comments can target different parts of a document. He extracts features from the comments using automatic extraction, which are the nouns, verbs, prepositions. Through the conversation graph, then can select passages by analyzing the parts of the conversation. Then he did a study to ask the relevant parts of the comments from students. I asked the question about what he used for extracting the nouns, verbs and prepositions from the comments, and he mentioned that he used lexical analyzers. Of course, this leads to the issue about spam comments, because some spam comments would pass the lexical analyzer.

The second presentation is Templates and Queries in Contextual Hypermedia presented by Frank Allan Hansen. To support contextual augmentation, there is a need to represent context in hypermedia. Context is represented in hypermedia by the HyCon implementation and using a UML diagram. Tagged objects can represent both data and context information, so need to look into structural computing for determining the structure behind the context. Structural templates are used to model the context and data which is borrowed from structural and object-oriented computing. Contexts are modeled as queries in HyCon. This is just a proof of concept model right now, so they haven't done evaluation of the contextual model.

The third presentation is Harvesting Social Knowledge from Folksonomies by Harris Wu. The research problem is whether social tagging can be used as a part of IT architecture. They are interested in determining how social tagging can be used as navigational links. This is a short paper and so there is a review of what is being done to finding social knowledge, like link analysis, data reduction using singular vector decomposition. It's interesting that he is basing his talk from other people's talks at the conference as examples of how other people are harvesting social knowledge.

The fourth and last presentation is Supporting the Design of Behaviours in Callimachus, presented by Manolis Tzagarakis. They view behaviours from a systems point of view. Within structures, they can determine propagation phenomena of operations. Phenomena-in-the large are expressed in terms of interactions-in-the-small. They use pattern-based approach with propagation templates.

This concludes the end of today!

Technorati tags: , ,

No comments: