中文摘要：

离散数据在Web网络中分布较广，是造成数据挖掘有用信息容量低的主要原因。霍金斯离散数据挖掘方法自提出以来获得了很高的成就，但仍存在挖掘数据分类性能不高的缺点，在此，使用BP神经网络对其进行改进。霍金斯离散数据挖掘方法分离散数据扫描和离散信息挖掘两个步骤进行，所提改进方法通过优化原方法中离散数据的排序规律，挖掘最优BP神经网络连接节点权值集群，改进离散数据集群的正确分区能力，降低离散信息挖掘过程的时空复杂度，提高原方法的分类精度和分类效率。实验结果表明，所提改进方法在Web网络离散数据中能获取高度可靠的挖掘结果。

英文摘要：

Discrete data has a wide distribution in the Web network, and is the main reason causing the low capacity of useful information. Since Hawkins discrete data mining method was put forward, it has been obtain a high achievement, but it still exists a fault that its data classification performance is not high. Therefore, the BP neural network is adopted to improve it. Hawkins discrete data mining method is divided into two steps： discrete data mining and diserete information mining. The im- proved method can optimize the discrete data sorting law of the original method, mine the optimal weight of BP neural network connecting node, improve the correct partition ability of discrete data cluster, reduce the time and space complexity in the pro- cess of discrete information mining, and improve the classification accuracy and classification efficiency of the original method. The experimental results show that the improved method can obtain highly reliable mining results in discrete data of Web net- work.