Constructing Researchers' Profile in Power domain Based on Knowledge Graph

This paper introduces an innovative approach to constructing researcher profiles within the electric power domain using knowledge graph technology. The proposed methodology consists of five key stages: data collection from multiple sources, data preprocessing, knowledge graph construction, profile labelling, and practical applications. The collected data encompasses demographic attributes, research outputs, project involvement, behavioral preferences, and collaboration information. Data preprocessing involves structuring this information via named entity recognition, utilizing advanced natural language processing tools. The knowledge graph is constructed using RDF triples, enabling data integration and minimizing redundancy through knowledge fusion techniques. Profile labelling is achieved through a combination of statistical, rule-based, and machine-learning methods, including the use of LDA models to extract research interests. The system is designed to update dynamically, incorporating new publications and modifications in researcher data, with a time decay factor applied to weigh contributions. The resulting profiles facilitate intelligent research collaboration, promote knowledge sharing, and minimize data silos. This structured approach enhances the identification of key contributors, supports cooperation, and streamlines research management within the power industry. The study highlights the significance of comprehensive data integration for efficient research profiling and informed decision-making.