Using Server Logfiles to Improve Website Design
Nowick, Elaine, Library Philosophy and Practice
To provide appropriate web-based information in an accessible form, librarians need to know how users visualize the organization of information within the system and how they search for the information they need. The Internet has been compared to a very large library before classification schemes were devised. The amount of information is incredible, but trying to find relevant answers to questions is a frustrating experience. The Agricultural Networked Information Center (AgNIC) is an effort by a coalition of land-grant universities and the National Agricultural Library (NAL) to provide a user-friendly guide to reliable information on the Internet on all aspects of agriculture. The University of Nebraska-Lincoln (UNL) joined AgNIC in September of 1996, providing a website serving as a clearinghouse to Internet sites useful for plant scientists and extension researchers. The UNL AgNIC Plant Science website is located at: www.unl.edu/agnicpls/agnic.html. The website consists of a main page listing topics within the field of plant science, such as crop statistical information, genetics information, and information on plant pests. These topics are then linked to subpages with lists of relevant websites. Short descriptions are included with each website listed. Another important feature of AgNIC allows users to access a reference service.
In developing UNL's plant science website, feedback and advice on website design was sought from other members of the AgNIC coalition and from plant scientists at UNL. Since this website is publicly accessible, most users are from outside of the UNL community. One way to learn more about these users is from the server logfiles.
Logfiles provide a record of communications between the server and computers requesting information residing on the server. When a user accesses a website, the browser program on the user's computer communicates with the server computer that hosts the website. The server sends the requested file to the user's computer and it then appears as a web page on the user's screen. The information coding each linked page is sent as a separate file. When the user downloads a file, the host server records the address of the computer to which the file was sent, the date and time of the request, and the name of the file sent. The logfile can be analyzed to tell something about the user, what subpages were viewed and in which order, which in turn indicates the navigation patterns of the user. Logfile analysis has been suggested as a way to understand user behavior in hypertext systems (Barab, 1996), to improve OPAC use efficiency (Connaway, et al. 1995), and to improve website design (Suleman, et al., 2000).
The purpose of this study was to identify ways that the design of the AgNIC Plant Science website could be improved by analyzing server logfiles to determine the types of users who accessed the site, the kinds of information they were seeking, and how they navigated within the site.
Logfiles for the UNL AgNIC Plant Science website were analyzed for two time periods: October 2, 1996-November 1, 1996 and December 11,1996-January 28, 1997 (Time 1) when the site first was published and July 30, 1997-October 2, 1997 (Time 2). These time periods were used because logfiles were available. The total number of users, the number of users per day, and the average number of users for each day of the week were examined for the two time periods.
A sample logfile entry is shown in Figure 1. The requesting computer is identified by an IP address. There may be one or more sets of numbers or letters at the beginning of the IP address that identify the specific computer to which the file will be sent. In the sample entry these numbers are represented by x's. The next set of letters (unl) indicates the institution that has registered the IP address, in this case the University of Nebraska-Lincoln. The last set of letters (edu) indicates the domain of the institution. …