Extracting Semantic Information from Patent Claims Using Phrasal Structure Annotations

Abstract

The rapid change of trading values from tangible assets to Intelectual Property has put both businesses and academia in a race to acquire and protect the rights to exploit such property. This is mainly accomplished in the form of patent issuing by the governments, being time consuming and complicated due to the vast amount of documents that need to be analyzed in order to assert the novelty or validity of a patent application. Patent information retrieval research is thus growing quickly to support document analysis across multiple domains and information systems. One of the big challenges in patent analysis is the identification of the elements of innovation (concepts, processes, materials) and the relations between them, in the patent text. This paper presents a method for extracting semantic information from patent claims by using semantic annotations on phrasal structures, abstracting domain ontology information and outputting ontology-friendly structures to achieve generalization. An extraction system built upon the method is briefly evaluated on a document sample from INPI, the Brazilian patent office, a challenging information source.