Using Interactive Visualizations of WWW Log Data to Characterize Access Patterns and Inform Site Design
The authors argue that most software packages only provide static visualisations of log file data. Due to the sheer quantity and complexity of these data insights through static visualisations are limited. The user must be able to filter and zoom the examined data interactively in order to gain more insights.
Hochheiser and Shneiderman begin with a brief overview of current efforts in log file visualisations. Modern software packages feature aggregation, filtering, and even analysis of individual user “trails". Unfortunately, these features often require a pre-processing step. Spotfire, however, allows the user to select the filtering criteria interactively using sliders. The main window of Spotfire provides a two-dimensional plot. Points in the plane can display more values by using size and colour coding. The authors then show six examples of plots that enable web site designers and operators answer relevant questions. The colour coding is extremely helpful to detect errors like broken links or accidentally deleted documents. The two-dimensional plot provides insights at a glance. For example, a URL vs. time plot shows which documents are retrieved at what time. Vertical clusters indicate a crawling/spidering activity. Each of the six plots is examined in detail and various effects are explained.
There are limitations that appear in most log file analysis software: It is very difficult to identify individual web users. For example, some internet service providers prevent host name resolution, others use proxy servers. Sometimes several persons use the same internet terminal, e. g. in a library or an internet café. Another problem is referrer spam or malicious crawling/spidering of web sites. When trying to answer web design questions with respect to usability aspects one has to distinguish between human users and web robots, and individual users must be identifiable. Unfortunately, Spotfire does not provide a mechanism to support either of these differentiations.
The tool presented by Hochheiser and Shneiderman is a very interesting piece of software because it allows user to explore log file data interactively. Some questions come only to the users’ mind when “playing around" with different parameters. An important aspect of log file analysis is the level of granularity. Unlike pre-processing tools Spotfire allows to adjust this level for each parameter interactively. The article closes with potential feature for future implementations. For example real-time monitoring of log files might prove useful for web site operators. The article presents many ideas of interpreting log file data. It is worth reading for web site designers and operators. Even if you are not considering using Spotfire the article provides you with insights that might change the way you look at log files.
Dieser Text erwähnt...
KB IB clear
|HCI/MMI (Human-Computer-Interaction)Human-Computer-Interaction, Informationinformation, Logfile-Auswertung, Softwaresoftware, Visualisierungvisualization, Webdesignweb design, WWW (World Wide Web)World Wide Web|
|Verweise auf diesen Text|
|Verweise von diesem Text|
|Webzugriffe auf diesen Text|
Falls Ihnen diese Seite gefallen hat
Besucher(06.17): 000008 *
Besucher Total : 000499 *
Beats Biblionetz ist Teilnehmer des Partnerprogramms von Amazon Europe S.à.r.l.
Beat Döbeli Honegger ist bei Google+
*(ohne Suchmaschinen und ohne Proxy-Verluste)