Abstract: This paper analyzed the necessity for the construction and implementation of self-organizing model of scientific data set in university library in the era of big data, and summarized the construction principles. Based on the discussion about the main difficulties of self-organizing model construction, it proposed the implement strategies from computing system security management, library data resource integration, data management and maintenance standards, resources application platform sharing service platform, to improve the data processing efficiency with the function extending as the core. Combining with the real-time, interactive and sharing characteristics of self-organizing model data resources of scientific data set in university library, it discussed the advancement, reliability, expansibility of self-organizing model, and proposed the system construction methods with computer, big data technologies as the core.