The purpose of this paper is to effectively evaluate the innovation ability and classification of technical talents in the intelligent robot field, and to be able to carry out adaptive learning and mining technical innovation talents according to the real-time change data corresponding to different indicators. Taking inventor’s patent information retrieved and cleaned from DI database as research object, it constructs the evaluation index system of technological innovation talents. It reduces the dimension of the index and cluster automatically, shows the visual effect, and mines the similar technical innovation talents through t-SNE algorithm. For a large number of patent information data, machine learning algorithm improves the traditional recognition method. According to inventor similarity, automatic classification is realized. Combined with DWPI manual code mining, the corresponding innovators and members of the technical team in the intelligent robot technology field were found. According to the results of visual dimensional reduction, the specific inventors can be traced. Machine learning algorithm t-SNE can reduce dimension and analysis clustering. It breaks the limitations of artificial statistics, deals with the larger order of magnitude data, and analyzes data timely, accurate and intuitive.
Author(s): Ning Zhao, Guohui Yang, Yang Cao
Organization(s): Harbin Institute of Technology
Source: 2020 IEEE International Conference on Artificial Intelligence and Computer Applications (ICAICA)