Conclusions

For many businesses, the trail of data left by users visiting the company Website is an untapped resource. Each visit to the Website generates several lines in the Web log file, each of which has only a few pieces of relevant information. New tools are actively being developed to help Web professionals interpret this large, complex data set. The present research explores the application of a new tool, the self-organizing map neural network to the problem of Web usage mining.

The visualization of the self-organizing map presented here is unlike earlier work because it enables the interpretation of Website usage by individuals who do not have extensive training. Earlier Web usage mining procedures automate the analysis so that no human is involved (Mobasher et al., 2000), or use special data mining languages for interacting with the system (Spiliopoulou, 2000). By using the natural grouping of the self-organizing map, the present visualization allows an inexperienced individual to identify usage clusters, and to understand which pages are visited together frequently. Such a simple, specialized tool should enable a wide variety of practitioners to gain a better understanding of their Website users.