摘要
使用用户所提供的博客特征信息,及广告点击反馈记录,形成训练集数据,运用改进的决策树算法生成的分类规则,进行数据挖掘分析,判断对特定用户和同一类型广告可能的投放效果,为广告投放提供可靠的依据。同时针对广告匹配的特点,对ID3决策树中偏向于多值属性进行改进,转换多值属性化为二值或三值属性,在数据预处理方面,提出了适合连续数据离散化的方法。从反馈的点击数据得到验证,此方法对提高广告的有效点击次数有明显的效果。
The character information of the blog s users and the feedback from the ad-clicks were used to form the set of training data. According to the data mining method,it generated the decision tree. The decision tree was employed to classify the effect of advertisement. The rules of classification guided the platform how to advertise the some type of the advertisement in special users blog. According to the character of the platform,a special improved ID3 algorithm was used to improve the bias of the multi-valu...
出处
《微型电脑应用》
2010年第5期4-5,8+72,共4页
Microcomputer Applications
关键词
广告平台
数据挖掘
决策树
广告投放效果
the Platform of Advertisement
Data Mining
Decision Tree
the Effect of Advertisement
