Abstract
In the highly competitive business environment, companies use patents as the intellectual asset to gain strategic competiveness. Patent prior art retrieval is a nontrivial task for invalidity and patentability search. This study adopts a hybrid approach to include claim structure and patent description to enhance prior art retrieval performance in terms of recall rate. We used USPC class 438 as the data set, which contains 52,311 patents from 1976 to 2005, and we randomly selected 300 patents in 2005 as the testing data. To evaluate the proposed hybrid approach, we designed a series of experiments; first, we identified the best parameter settings in claim structure; second, we decide the best ratios of sentences in description as relevant contents; finally, we obtain the best result by combining the information of claim structure and top 70% sentences in description. We successfully achieve the competitive result in terms of raising the recall rate with the proposed hybrid approach, which also demonstrates the usefulness of including claim structure into prior art retrieval system.