Chanda Harris

Zoomable User Interfaces

CMSC 838B

February 21, 1999

The paper, "Graphical Multiscale Web Histories: A Study of PadPrints", discusses a browser tool called PadPrints developed to aid in Web navigation, particularly accessing previously visited pages. PadPrints uses a zooming user interface (ZUI) called Pad++ to display Web page history and is considered a Pad++ application. Web page history is displayed in a separate window from the browser, hierarchically according to the structure of a Web site. Pages already present in the hierarchy are not duplicated. Clicking on a page displays the corresponding page in the Web browser. The Tcl scripting language was used to develop PadPrints. PadPrints works along with a traditional Web browser, requiring some reconfiguration to the cache and proxy server. PadPrints, with its zooming multiscale user interface, is an approach in that tries to solve the problem of graphically displaying large quantities of nodes in a minimal space.

Two usability studies were performed to test the effectiveness of PadPrints based on the hypothesis that a graphical hierarchical history tool would result in an improvement in the speed and number of pages accessed in comparison to just having the traditional history mechanisms, particularly the "Back" and "Forward" buttons, the "Bookmark" list, and the "Go" button in Netscape Navigator 3.0 which was the comparison browser used in the studies. It was also hypothesized that PadPrints would be more favored than using Netscape Navigator 3.0 alone.

The first study used a set of pages organized hierarchically with no images or graphics in order to test simple navigation. The second study used pages, not entirely hierarchical, containing images. For both studies, the task was to locate pages and answer questions about particular pages. The second study specifically tested the time it took, using PadPrints, to visit previously accessed pages. Overall, there was no significant difference in either study in the time to answer questions between subjects using PadPrints and those using Netscape Navigator 3.0. However, studies did reveal a significant difference (a decrease) in the time to answer questions using PadPrints for revisiting previously accessed pages.

WebMap displays the relationship between pages by using a small circle to represent each page that, when selected, displays the actual page. It uses color-coding to provide information about links.

http://www.tm.informatik.uni-frankfurt.de/~doemel/Papers/WWWFall94/www-fall94.html

WebBook is a three-dimensional browser that supports viewing multiple pages simultaneously and allows the storing of pages in books rather than bookmark files.

http://www.acm.org/sigs/sigchi/chi96/proceedings/papers/Card/skc1txt.html

MosaicG is very similar to PadPrints in that it has a graphical history mechanism. It constructs maps as users visit pages, resulting in a map of web usage.

Overall, I found that the paper was very well-written and readable and that it presented PadPrints in a very understandable way. PadPrints does indeed seem to be a significant contribution in browsing Web page history. However, I am not convinced that the experiments performed to test the effectiveness of PadPrints really tested the motivation for developing PadPrints, displaying large numbers of nodes in a minimal space. Also, there is no mention of the number of pages contained in the sets of Web pages used in the experiments. This would be useful to know in assessing the effectiveness of PadPrints in revisiting pages when lots of pages have been accessed (i.e. the size of the images would get smaller and smaller as the number of pages accessed increased), and this would likely hinder a user’s ability to identify previously visited pages, and thus take more time. It probably would have been more effective and useful to display all statistics graphically and provide more details used in the experiment.

Some questions I had as a result of reading this paper are listed below.

1. Does the user have any control over the zooming?

2. Were there any observation about zooming that resulted from performing the experiments?

3. Can Pad++ be run on other operating systems?

4. Was there any affect in the results caused by using five of the subjects for both experiments?

5. How does PadPrints scale on revisiting pages when the hierarchy consists of a large quantity of nodes?

I do think that this work is definitely a significant and unique contribution to help users navigate through their navigation history. It also has the potential to help users understand the structure of any given Website.