ABSTRACT
The problem that inspired this volume is how to develop intelligent computer systems that are text-based—systems that acquire their knowledge by assimilating massive amounts of ordinary natural language text, rather than having to be spoon-fed rules handcrafted by knowledge engineers. 1 A mature text-based system would keep up with current events by simply reading the newspaper as we do. Conceivably, it might even learn new fields by reading the textbooks.
