BUILDING A KNOWLEDGE BASE TAXONOMY FROM STRUCTURED OR UNSTRUCTURED COMPUTER TEXT FOR USE IN AUTOMATED USER INTERACTIONS