Abstract

Abstract: ver a year ago, a small research project aiming at extracting causality information from Root Cause Analysis (RCA) documents was started. This is a good example of acquiring information from legacy documents using Natural Language Processing, Ontology and Information Retrieval techniques. Causality expressions are good examples of understanding how semantic annotations could be made automatically exploiting a small number of linguistic patterns that are obtained from a corpus analysis.

In this talk, challenges of reusing RCA findings and motivations of applying automatic extractions are discussed. Details of the proposed approach are presented including current status of the project and a plan of user evaluation.