Pattern Identification and Correction of Document Misinterpretations in a Natural Language Processing System

An approach is provided in which a knowledge manager analyzes multiple document phrases using a natural language processing model and generates multiple interpretations based upon the analysis. The knowledge manager identifies misinterpretation patterns by comparing the multiple interpretations with multiple corrections that include corrections to the multiple interpretations. In turn, the knowledge manager generates interpretation rules based upon the identified patterns and applies the interpretation rules to the natural language processing model.