论文部分内容阅读
为改善用户的Web页面访问行为、提高访问效率,设计了一种基于贝叶斯网络的网页推荐模型及推荐算法。通过收集和分析服务器中的描述文件和日志文件,利用Bayesian网络分析页面间的依赖关系,构建了基于贝叶斯网络的网页推荐模型并产生推荐集。通过在Microsoft公司提供的网络日志数据集上做的实验,可以获得超过80%的准确率和覆盖率。理论分析和实验结果表明:算法能够在线实时向用户做出个性化的推荐,与已有的推荐算法相比,算法能较快地给出推荐集,并且可以获得更高的准确率和覆盖率。
In order to improve the user’s Web page access behavior and improve the access efficiency, a Web page recommendation model and a recommendation algorithm based on Bayesian network are designed. By collecting and analyzing the description files and log files in the server and using Bayesian network to analyze the dependencies between pages, a Web page recommendation model based on Bayesian network is built and a recommendation set is generated. By experimenting with Weblog data sets from Microsoft, you get over 80% accuracy and coverage. Theoretical analysis and experimental results show that the algorithm can make personalized recommendation to users online in real time. Compared with the existing recommended algorithms, the algorithm can give a recommendation set faster, and can get higher accuracy and coverage .