数据挖掘方法的评述

Methods for data mining

决策离不开知识 ,从数据库中采掘知识 ,是解决从大信息量中获取有用知识的有效途径。但是在实际数据库中 ,数据的复杂性 (如信息量大、噪声等 )对数据挖掘方法提出了比机器学习更高的要求 ,这方面的研究正受到越来越多的关注。本文就当前数据挖掘的几种主要方法 ,即神经网络、决策树、粗集和云模型等方法的研究现状进行了评述 ,指出其存在的问题。从总体上看 ,这些方法都有局限性 ,但它们的有机组合具有互补性 ,多方法融合将成为数据挖掘的发展趋势 。

Knowledge is vital to decision. In order to solve the problem 'We are drowning in data, but starving for information', data mining from database(DMDB) is necessary for the non trivial extraction of implicit,previously unknown,and potentially useful information. DMDB is more difficult than machine learning because large number of small,very specific and full of noise instances are dealt with. Researchers in KDD pay more attention to developing and implementing efficient data mining techniques. Some main data mining methods ,for example, decision trees, neural networks, rough sets, and cloud model, are reviewed in detail and the difficulty existing in them and research focuses in the future are pointed out in this paper. From the higher level, the methods are used limited to a certain circumstances but their integration can improve the situation.