We present a general framework for the information extraction from web page
s based on a special wrapper language, called token-templates. By using tok
en-templates in conjunction with logic programs we are able to reason about
web page contents, search and collect facts and derive new facts from vari
ous web pages. We give a formal definition for the semantics of logic progr
ams extended by token-templates and define a general answer-complete calcul
us for these extended programs. These methods and techniques are used to bu
ild intelligent mediators and web information systems.