一种适用于云图书馆系统架构的自适应洗牌方法

An Adaptive Shuffling for Cloud Library System Architecture

提出了一种自适应重采样方法,即自适应洗牌方法。它可以自适应的调整包的大小。该方法解决了当新包过大则只有少量的新包被加入到训练数据中且包越大向分类器提供的信息也就越少,结果对生成的分类器影响很小;若新包过小则容易生成错误的包而在训练集中加入噪声等问题。实验结果表明,本文方法可以显著提高分类器的准确度。

With the promotion of the cloud library, a high accuracy book recomendation system is essential to borrow more efficiently for reader. The adaptive resampling method （i. e. Adaptive Shuffling） is proposed to make the most use of known information and achieve a better accuracy. It can adaptively adjust the size of the bag. The experimental results show that the proposed method improves largely the accuracy of the classifier.