VISITaTOR

About

Most of today's log analyzers can show the number of times a certain page has been requested, how popular it is, and sometimes the clickstreams as well.
While this data is, no doubt, important, there is obviously much more to be mined in logfiles.

I have developed a method of grouping the site visitors according to the pages they request and clustering them step by step into increasingly common (high-level) groups.

• The uppermost level contains groups which have little or no elements in common, the lowermost - those which almost coincide.
  Thus, the entirety of access patterns is represented by a tree.

  One can draw conclusions regarding the interests of the visitors who make up a group from the content of the pages they visit.

  The logfiles to be analyzed should be taken from a stable period of time. You should exclude accesses by robots and non-informative accesses, like images, stylesheets, subframes etc.

  If the site navigation occurs dynamically, using parameters, the current version of VISITaTOR program will probably not be of much help to you, because parameters are ignored in it for the purpose of simplification.

  It is advisable to analyze a few series of data from various time periods (the site content for all of which should be the same), and compare the results.
  If you encounter the same groups, there is reason to belive that they represent reality; if a group is instable, this group is probably accidental.

  The more data you have, the more dependable the results are.

Therefore I am interested in proposals for developing this tool, which is rather imperfect so far, into a quality product.