Abstract
Paten retrieval plays a very important role for research and development teams to design new products. Product managers must retrieve relevant patents efficiently and correctly before developing products in order to verify the competitive advantage of new products. Enterprises could prevent from loosing of tangible and intangible assets for patent infringements by retrieving potential relevant patents. Domain knowledge is indispensable while searching relevant patents in an effective and efficient way. If product mangers could retrieve relevant patent which are important and relevant to the new developing product by their own basic technical knowledge in the field, it could save the cost of time and human resources. A bootstrapping patent retrieval approach proposed by Chen (2008) provides a mechanism which could expand the patents identified by experts into a scope including possible relevant patent sets. It reduces the effort of computation for comparing patent similarity with the structure of inventions. In this research, we proposed a bootstrapping approach modified from Chen’s approach (Chen, 2008), and verified whether the patents identified by domain experts could be replaced or not. That is, is the bootstrapping patent retrieval approach sensible to the input patents? If a product manager could expand the patent into a patent set similar to what experts’ have expanded by his technical domain knowledge, the product manager could save the cost of retrieving patents to the relevant target products. The evaluation results show that the bootstrapping patent retrieval system is sensible to the input patents; that is, the candidate patent sets expanded by patents suggested by experts vs. non-experts differ significantly.