Improving University Faculty Evaluations via multi-view Knowledge Graph

University faculties generate a large amount of heterogeneous data in e-learning environments that online systems and toolkits have made widely available in all aspects of teaching and scientific researching activities. How to use the data efficiently and scientifically for faculty evaluations has recently become an important issue in university performance systems. However, it is still a challenge to comprehensively assess faculty members using multi-source and multi-modal data due to the lack of uniform representations and evaluation processes. To this end, this paper proposes a novel University Faculty Evaluation System based on a multi-view Knowledge Graph (UFES-KG) that integrates heterogeneous faculty data. Relevant data, collected both on the Internet and through university-administered internal systems, includes faculty information such as scientific research papers, patents, funds, monographs, awards, professional activities and teaching performance. Furthermore, we construct entity representations through knowledge graph embedding methods to retain their semantic information. In addition, by integrating the academic development status of scholars in the previous three years as well as student evaluation data, this paper proposes an academic development factor (ADF) for making predictions about faculty academic development. The experimental results show that this factor is closely related to the features of the knowledge graph and student evaluations. In a certain case study, this factor is superior to the traditional h-index, g-index, and RG score. Intuitively and scientifically, this multi-view approach can improve evaluations of university faculties.