A User Requirement Feature Similarity-Based Computing Power Resource Matching Method and System

abstract-Computing power network needs to manage, allocate and schedule various computing power resources. The key objective is to optimally allocate computing resources based on user demand. Addressing the challenge of matching computing resources to diverse user requirements, we propose a user requirement feature similarity-based computing power resource matching method and system. This method combines user requirement features, historical user allocation results of computing power resources, and user evaluations to achieve optimal matching of computing resources. By matching current with historical users based on feature similarity, it gathers similar users’ resource allocation information and conducts further screening based on user evaluations. Its simplicity allows for quick and effective selection of suitable computing resources, enhancing matching efficiency and accuracy. In our experiments, various scenarios were simulated, and the method described here was used along with delay-first and price-first greedy algorithms to match users with computing resources. To evaluate the effectiveness of the proposed method, we utilized indicators such as request processing delay, unit price, and the frequency of user evaluation levels. Notably, using the proposed method, 81.46% of the samples rated “excellent” significantly outperformed the comparison algorithm. Our approach comprehensively addresses both request processing delay and the cost of using computing resources, showcasing competitiveness in both these crucial indicators.