Visual Thesaurus is based on wikipedia.org and Google images search. The articles of the english Wikipedia are building the vocabulary of the thesaurus. The terms are fed to the Google images search. The first hit of the search results and the title of the article represent the concept. Around this concept, thematically related terms are shown. The terms act as hyperlinks to new concepts. This way, an associative word-image-net (with sometimes iconic character) is evolving.
Visual Thesaurus can be useful as a research tool, but can be explored as a sort of hyper fictional construct aswell.
This project is possible through a webservice offered by wikipedia-lab.