排基于机器学习的电子资源评价方法研究

Purpose/Significance : Establishing a scientific and efficient electronic resource evaluation method is an important task for the construction of university library resources and the improvement of service quality. Compared with traditional electronic resource evaluation methods, machine learning models provide more accurate and efficient evaluation and prediction results, and guide more objective optimization of electronic resource structure. [Method/Process]: Taking foreign electronic resources as the research object, establish two different types of evaluation index systems: foreign journal full-text library and abstract library. Collect DRAA website single library statistical reports and evaluation center data, apply RF (Random Forest) to rank the importance of feature variables, compare and construct machine learning models, use leave one cross validation to predict electronic resource evaluation, and compare the accuracy to ensure the optimal performance of the model. Result/Conclusion : Machine learning models are more accurate and efficient in predicting the evaluation of electronic resources than traditional methods. They can make data-driven decisions for libraries and help optimize the allocation of electronic resources.