Recommendation of collaborative filtering for a technological surveillance model using Multi-Dimension Tensor Factorization

Abstract

Technological surveillance in research centers and universities focuses on carrying out a systematic follow-up on the development of research lines, the research leaders, the possibilities of scientific-technological collaboration, and to the knowledge of current trends from research. All these elements allow guiding the researches and supporting the scientific-technological strategy. This research proposes a model of technological surveillance supported by a recommendation system as an application that focuses on the preferences of researchers in universities and research centers. The multidimensional tensor factorization approach, based on grouping to build a recommendation system and to validate the increase in tensors, improves the accuracy of the recommendation. The experiments have been carried out in real data sets as the university and research centers. The results confirm that the dispersion issues are improved within a wider margin in both data sets. In addition, the proposed approach states that the increase in the number of dimensions shows a 7-10% improvement in accuracy and memory, which increases performance as an expert recommendation system.