Language related problems in the IPC and search systems using natural language

The value of the IPC as a search retrieval tool, independent of the specific language of both the documents retrieved and the user, is briefly reviewed. Emphasis is laid on the limitations of any form of word searching, especially when synonymous terms need to be found in more than one language in the course of a search. The need for the IPC to keep pace with both the developments of the technology of the inventions in patents and the rapid changes in the field of electronic retrieval of patent information, is highlighted. The practicalities and implications of the establishment of the authentic text in both English and French, with "official" texts and "unofficial" texts in several more languages, for the convenience of users, are explored. The process of translation must, at least in the translation into the French or English initial and amended authentic texts, exactly mirror the concepts expressed in the language in which the particular sections of the IPC or amendments were established. In this process errors and inconsistencies in the version in the first language are identified, and hence contribute further to the improvement of the IPC. A natural language system for interrogating the text of the IPC in French (CIB-LN), to enable users unfamiliar with the IPC to input their query in natural language and receive the most suitable IPC terms in reply, is also described. INPI has developed this system in response to user demand in searching the commercial and, more especially, the free Internet patent databases. The system has been designed to be adaptable to similar needs in other languages.