A method and apparatus for abstracting meanings from natural language words. Each word is analyzed for its semantic content by mapping into its category of meanings within each of four levels of abstraction. The preferred embodiment uses Roget's Thesaurus and Index of Classification to determine the levels of abstraction and category of meanings for words. Each of a sequence of words is mapped into the various levels of abstraction, forming a file of category of meanings for each of the words. The common categories between words are determined, and these common elements are output as data indicative of the most likely categories of meaning in each of the levels of abstraction to indicate the proper meaning intended to be conveyed. This data is then processed according to a rule to obtain a result.