Measuring Patent Similarity Based on Text Mining and Image Recognition

Patent application is one of the important ways to protect innovation achievements that have great commercial value for enterprises; it initial step enterprises set business development track, as well a powerful means their core competitiveness. The emergence large amount patent data makes effective detection difficult, and infringement cases occur frequently. Manual measurement in slow, costly, subjective, can only play an auxiliary role measuring validity patents. Protecting inventive holders realizing more accurate were issues explored by academics. There are five main methods measure similarity: clustering-based method, vector space model (VSM)-based subject–action–object (SAO) structure-based deep learning-based method. To solve this problem, paper proposes calculation method fuse similarity text image. Firstly, SAO structure extraction technique used obtain content text, compared similarity; secondly, image information extracted compared; finally, obtained fusing two aspects information. feasibility effectiveness scheme proven studying number field mechanical structures.