A Semi-Supervised Machine Learning Method for Chinese Patent Effect Annotation

Patents are public and scientific literatures protected by the law, and their abstracts highly contain valuable information. Patent's semantic annotation can effectively protect intellectual property rights and promote corporations' scientific research innovation. Currently, automatic patent annotation mainly uses supervised machine learning algorithms, which is required abundant expensive labeled patent data. Due to lack of enough labeled Chinese patent data, this paper adopts a semi-supervised machine learning method named co-training, which starts from a little labeled data. This method cooperates keyword extraction with list extraction, and incrementally annotates functional clauses in patent abstract. Experiment results indicate this method can gradually improve the recall without sacrificing too much precision.