Method of representing high-dimensional information

In a computer network environment, a method of clustering visitors to a web site, comprising:

• identifying web pages within said web site visited by said visitors;
• generating context vectors for said web pages;
• summarizing the generated context vectors to generate a summary context vector for said visitors;
• storing said summary context vectors in a database;
• identifying visitors with similar summary context vectors;
• clustering said identified visitors with similar summary context vectors;
• labeling said cluster of visitors;
• and
• storing information pertaining to each labeled cluster in said database.

In a computer network environment, a method of identifying clusters of visitors to a first web site, comprising:

• extracting domain information from said visitors;
• identifying at least one second web site associated with said visitors based on said extracted domain information;
• generating context vectors for each of said at least one second web sites, wherein said context vectors are representative of the content in said second web sites;
• clustering said generated context vectors so as identify clusters of visitors associated with said at least one second web site having similar content;
• labeling said cluster of visitors;
• and
• storing information pertaining to each labeled cluster in a database.