中文摘要：

数据分类是数据挖掘的主要内容之一，通过分析训练数据样本，产生关于类别的精确描述。贝叶斯分类是数据挖掘领域中一种常用的有效分类方法。在关系学习中，贝叶斯分类算法有很多种，对这些算法进行总结、比较，指出其优点与不足，对提高分类效率有很大帮助。本文对已有的关系学习中贝叶斯分类算法作了详细的比较，并进行归纳总结。在单关系学习中重点介绍了几种基于粗糙集的贝叶斯分类器和加权贝叶斯分类算法，并分析了各种方法的模型、权值确定方法、优缺点及进一步工作方向。在多关系学习中主要比较了几种基于语义关系图的贝叶斯分类算法，重点介绍了MI—MRNBC模型。最后对本文工作进行了总结与展望，提出进一步工作方向是研究基于粗糙集的多关系贝叶斯分类算法。

英文摘要：

Data classification is one of the main content of data mining. Through analyzing training data samples, it is resulted in the accurate description on the classification. Bayesian classification is an effective simple classification algorithm in the field of data mining. In the relational learning, there are many kinds of Bayesian classification algorithms. It would be of considerable help to improve classification efficiency that summarize, compare these algorithms and point out its advantages and disadvantages. In this paper, some algorithms have made a detailed comparison and summary. In single relational learning, it is focus on several Bayesian classification algorithms based on Rough set and weighted Bayesian classification algorithms and analysis the models, methods to determine weights, advantages, disadvantages and the direction of further work. In multi- relational learning, the main comparison is several kinds of Bayesian classification algorithms based on Semantic relationship graph and focuses on the MI-MRNBC model. Finally, it is the summary and prospect of this article. The direction of further work is to study multi-relational Bayesian classification algorithm based on Rough set.