Metaphor is a fundamentally knowledge-hungry phenomenon. We need to possess knowledge about a topic (that which is described in a metaphor) and a vehicle (that which does the describing) before we can meaningfully describe one in terms of the other. Where does a computer acquire this conventional knowledge? From conventional language, which is abundant on the World Wide Web.
So, to use metaphor effectively, one must engage in a process of creative introspection. That is, one must be capable of asking the right questions about a topic before deciding which metaphor is most apt, or deciding which aspects of a vehicle are most salient to our thinking about a topic.
A computer can acquire common-sense knowledge by looking at the questions that are commonly posed on the Web. The question "why do dogs wag their tails" tells a computer that dogs have tails and that they wag these tails. A computer does not need to answer these questions to find them informative: it simply needs to strip out the shared presuppositions on which the questions are based.
Metaphorize is an application that uses large amounts of common-sense knowledge that have been harvested from the Web in this fashion. Enter a topic like clown and Metaphorize will show you the knowledge it possesses, and present you with potential metaphors for viewing this topic through fresh Metaphor Eyes.
Enter a perspective like scientists as artists or artists as rebels or philosophers as poets and Metaphorize will show you which aspects of the vehicle can meaningfully be transferred to the topic.
Enter a lexicalized metaphor like war criminal or media empire or media circus and Metaphorize will show you which aspects of criminal or empire or circus appear to be exploited by the metaphor.
Metaphorize shows that to think metaphorically, a computer must be able to pose its own introspective questions. How better can a computer learn to introspect than by observing the questions asked by others on the Web?
Interact with Metaphorize now ...