Predicting the Legal Risk of "Section 337 Investigations" by Elastic Time Predictor

Abstract

In recent years, more and more patent lawsuits have been filed by Chinese enterprises, represented by the "Section 337 investigations" of the United States. In order to help Chinese enterprises cope with the challenges of patent litigation, a matrix factorization based recommendation system are used to predict the legal risk of 337 investigation. However, the results predicted by the model are prone to over-fitting. In order to solve this problem, this paper proposes a new recommendation framework, namely elastic time predictor. The model is a hybrid model combining matrix factorization and truncation function. We encode the information of the prosecution case of major companies and decompose it into two sub-matrices, and then combine the decomposed matrix with the segmentation of the truncation function to maintain the entire recommended frame flexible. In the recommended approach, we consider the risk of litigation that a company may experience when entering a new market, for example the risk that a potential competitor will file a lawsuit against a new entrant. We use actual data to conduct experiments, and the experimental results show that the proposed method is superior to the baseline method and has significant advantages.