First session: Information Searching

The first paper from the conference program is on Is A Good Title? This paper looks into is there a way to name a URI to give a better title that will help in retrieval of search results. Many times, when you enter a search term, the URIs provided are not the relevant results. The author mentions about title evolution and gave the example of Sun.com. How much do titles change over time? They analyzed their title performance prediction but didn't find any new evidence that a page of a page that has a title of more than 24 terms is most likely spam.

The second paper is on Parallel Browsing on the Web by Jeff Huang and Ryan White. Browsers have tabbed browsing and they look into parallel browsing and gathered data from Internet Explorer 8. This was joint work with Microsoft Research. From their results, they found that users don't visit more pages when using more tabs.

The third paper is on A semiotic approach for the generation of themed photo narratives by Charlie Hargood, David Millard, and Mark Weal. According to Charlie, narratives surround us and we need to create systems to accommodate these narratives. They take a semiotic approach where semiotics are the study of signs and how we understand them. They did an experiment to evaluate the effectiveness of the Themed Montage Builder. The Themed Montage Builder (TMB) is a prototype using definitions in terms of model to generate themed montages, eg. from Flickr images. They found that the TMB shows a slight improvement. It is interesting how this approach could be used to create narrative from travel blogs.

The last paper in this session is The Impact of Bookmarks and Annotations on Refinding Information by Ricardo Kawase, George Papadakis, Eelco Herder, and Wolfgang Nejdl. They created an annotation system called SpreadCrumbs which is a social annotation system and it supports hypertrails which is a path of links. They then did experiments to test their SpreadCrumbs system compared to Delicious for bookmarks and annotations. The results showed that it was faster for the group that used annotations and SpreadCrumbs to find the information and outperform search in terms of performance.