Classification of web tools

Reading a diploma work "Towards an Integrated Approach to Collaborative Web Usage" I have discovered an interesting classification of web tools created by following another article "Discovering User Access Patterns on the World-Wide Web". I citate this text below having done some minor additions and changes.

Level 0
A software system that “retrieves documents for a user under straight orders.”: the user must give the document's URI to the browser so that it can retrieve the document. The common term for this level tools is "Web browser". Notice however, that most currently available Web browsers extend the behavior at least by history and bookmark mechanisms.

Level 1
These tools provide “a user-initiated searching facility for finding relevant Web pages”. The most common example are Internet search engines. Current Web browsers often integrate search engines into their interface.

Level 2
Software systems that “maintain user profiles and have an active component for notifying users whenever new relevant information is found”. The user profiles in this class of Web tools are usually static: the user enters his interests and the system looks for information matching those interests (actually the same key words that he is using for level 1 tools). One real example of a tool from this category is Google alerts.

Level 3
A more dynamic and deductive approach qualifies a level 3 Web tool. While for a level 2 Web tool the user needs to be aware of his interests and must be capable of expressing them to the tool, level 3 Web tools attempt to infer the user profile by analyzing the user's behavior. This becomes particularly important as humans are not used to, and usually not capable of formalizing their browsing behavior or information needs because this is not needed in most every day situations. There are a lot of well known example of such systems, like for example Amazon or Google subsystems offering extra things (books, links) basing on (purchases, searches) history.

Level 4
A tool from this category should have “the capability of learning the behavior of both information users and information sources”.