A Trust-Enhanced Patent Recommendation Approach to University-Industry Technology Transfer

Abstract

Technology transfer enables the technology from legal owners to be used by others, and it is essential for technology innovation in modern society. However, transferring technology from academia to industry has become a challenging task due to the "cultural divide" problem, where researchers in universities tend to focus on knowledge discovery, while companies focus on making profits with application of proven technologies. This creates a mistrust problem for companies to use academic patents invented by universities. Various recommendation methods have been proposed for technology transfer purposes, but few have addressed the trust issue caused by the cultural divide. This paper proposes a multidimensional trust-enhanced recommendation approach to promote academic patent trading. The approach extracts patent information, users' online interactions, and technology transfer information for recommendation calculation. It includes 1) measuring the degree of connectivity between companies and patents by the Personalized PageRank model; 2) measuring the trustworthiness of a potential patent transaction from the aspects of patent quality, inventor, and university; and 3) adopting a logistic regression model to integrate the above measurements. The results of our user-based experiment show that the proposed recommendation approach obtains higher average hit rate and higher willingness scores than current recommendation methods.