Session 1: Blogs, Wikis & RSS
The first talk in the first session is CUTS: CUrvature-Based Development Pattern Analysis and Segmentation for Blogs and Other Text Streams by Yan Qi and K. Selcuk Candan. Their challenge is to extract information to enable indexing, mining and ease of navigation. They visualize a blog archive by using length of segments and gradient of segments to indicate log vs., short, more vs. less change and the degree of concentration (high vs. low). Their claim is that topic-based segments can enable blog search. Some people use curve sgementation in order to detect topic shift detection and measure teh similarity between consecutive entries. The topic shift is assumed to occur at local minima of the curve. The method is to take the blog entries and represent them as a vector, calculate similarity value among entry vectors and then construct a dissimilarity matrix. Then the entries in the matrix are mapped to a 2-D curve using multidimensional scaling based on entry dissimilarities. From here, they can figure out topic development patterns. To automatically determine the topic development patterns, they use the adaptive curve segmentation algorithm. They tested their algorithm with a blog archive and book.
The next speaker is me and I'm going to present now. I just finished presenting. I think it went well and I had lots of questions which I recorded so I can take that to improve my work. Thanks to this guy from the conference for taking pictures of me, he has a much better camera than mine!
I'm presenting
Marketing myself with URLs (hey, it's a Hypertext conference!)
My talk
The fourth speaker is by David Millard called Web 2.0: Hypertext by Any Other Name? I guess there is no third speaker apparently. This is based upon his Masters project. The original hypertext pioneers had objectives: 1) Halasz' seven issues, 2) open hypermedia systems, 3) adaptive hypermedia systems. In the work, he selected example Web 2.0 systems to determine if they conform to the original hypertext objectives. From the analysis, they created a table to compare the various sites and the features like content search, context search, structural search, dynamic content, links, versioning, annotation, personalization and extensibility to see if the sites did support this. It is important to note that none of the Web 2.0 systems have support for typed n-ary links, but the research systems do. Dynamic structures are also not prelevant, and he mentioned that (that's why my work could be applied, the dynamic social structures can be used to find community). Trails are also not supported, trying to figure out what the user did to navigate (we do have trackbacks in blogs, and search engines do crawling and caching, so it DEFINITELY should be possible to do this, but it's not being done automatically). That would be cool, to find a social trail behind the hypertext. In summary, most of the important aspirations of the hypertext community have been fulfilled in Web 2.0. One of the questions that I asked is that it's not really a fair assessment to compare the early hypertext pioneers because they didn't take into account the social collaboration and the open communication and user driven interface of the web now. So I proposed a new term, instead of hypertext, it should be called hyperspace.
Yahoo tags:
blogs, wikis, RSS, web2.0
Social Networks, Networking and Virtual Communities
It's after the coffee break, and the second session is on Social Networks, Networking and Virtual Communities. The first presenter is Cameron Marlow from Yahoo Research and their paper is on HT06, Tagging Paper, Taxonomy, Flickr, Academic Article, ToRead. This is a position paper and trying to classify the space of tagging. Tagging systems add users into the resources, whereas before, the users were experts and implicit in the web resources. They are creating a tagging model and create a taxonomy based on that, followed with a preliminary study. The systems taxonomy is looking into the structure of tags. Tags can be distinguished by permissions as to who can tag, as well as recommendation of tags. Another classification of tags is tag aggregation like the concept of a set in Flickr, and the object type that is being tagged like web pages, events, e-mail or photos, etc. The statistics in the number of distinct tags over time for a particular user seem to be similar between Flickr and del.icio.us. Flickr and del.icio.us are two completely different tagging systems. I asked the question about tag clouds, and whether Yahoo Research is looking into the analysis of tag clouds and how tags can be aggregated into a tag cloud, where certain tags are related to exactly the same thing (like for instance I tag this post with hypertext2006, hypertext06, ht06, ht2006, but they all mean the same thing). Yahoo is now providing data for academics to make use for.
Now, there is the presentation of short papers. The second presentation in this session is Social Navigation in Web Lectures being presented by Robert Mertens. The talk is about creating a web lecture interface. Social navigation is being used for e-learning.
There is no third presentation, so now there is the fourth presentation which is Using String-matching to Analyze Hypertext Navigation being presented by Roy Ruddle. The string-matching method looks at repeated subsequences in sessions where each letter in the subsequence is the link. One of the things that resonates with me is to provide a trail network (artificial paths) that is generated automatically. Using this type of analysis of navigation can be used to reconstruct sessions. Several people argued because we already have caching of web links (ie. Google personalized history), we already have trails of our navigation.
The last presentation before lunch (yay!) is A Cognitive and Social Framework for Shared Understanding in Cooperative Hypermedia Authoring by Weigang Wang. He is using Piaget's social cognitive theory and applying it to a shared hypermedia workspace.
Technorati tags: hypertext2006, hypertext06, ht2006, ht06
No comments:
Post a Comment