Paper Reading #18: Automatic Generation of Research Trails in Web History

Comments
Jorge Perez
Adam Friedli

Reference Information
Title: Automatic Generation of Research Trails in Web History
Authors: Elin Ronby Pederson, Karl Gyllstrom, Shengyin Go, Peter Jin Hong
When/Where: IUI 2010


Summary
A research trail is an ordered sequence of web pages, accessed as part of a larger search. They are often created as a researcher looks for information online. An ethnographic study was conducted, showing that many people are engaged in extensive research using the internet, but have problems maintaining context. Therefore, this paper proposes automatically generating research trails to help web users create context and to remember where they left off when conducting fragmented web research.

A real trail, since a research trail is a little harder to picture. Source: awtec.net

These research trails are constructed by evaluating users' history, using both semantic (focusing on the material the user interacts with) and activity (focusing on how a user interacts with data) analysis to group similar web pages together. The concept of early research is considered as an area that needs such trails, considering its fragmented nature and the occurrence of topic sliding, where the first step in the trail differs from the last step depending on the information gathered between them.  Using research trails, the user would be shown where he or she is currently at within a particular research endeavor.

Finally, the paper outlined an implementation of automatic generation of research trails within an internet browser. It captures users' history, detects topics, and clusters the topics into research trails using Google history data. In the new tab page, the user can see a list of the most recent research trails, and can request other research trails. These can each be viewed. A study was conducted with this system on three users, which confirmed the goals of the system.

Discussion 
I thought the idea of research trails was very interesting, and that it would be very useful to have something like the paper discussed, whereby research trails are recorded and available for viewing within the browser. It would have been nice to read more about the evaluation of the system, since only three people were included in the study and none of their opinions about the system were provided. However, the authors did state that they plan to conduct a more extensive evaluation in the future. I would love to be able to try this system out. I often find myself researching a particular topic, going from page to page on Google, when something that I found makes something discovered earlier make sense but I can no longer find that previous page. In addition, if I were to close out the browser and come back to later, I have generally forgotten where I was at. With research trails, all of this information would be easily accessible, making researching a topic much easier.