A lexico-syntactic Pattern Matching Method to Extract Idm- Triz Knowledge from On-line Patent Databases

Patents are increasingly of great importance to innovation performance. A patent document is an important source of technological knowledge where artifact evolves according to problems and partial solutions. This paper reports on an on-going research whose goal is to provide design engineer with an efficient patent mining tool that re-structures a patent corpus in a problem graph to better highlight state of the art of given field of knowledge. The methodology employed to achieve such a functionality is based on a NLP method that manage both IDM-TRIZ concepts and patent corpuses contents. The paper also presents a case study in steel making industry to illustrate the methodology.