基于用户行为分析的本地搜索排序算法优化

随着本地搜索的发展，通用排序算法得出的排序结果已不能完全满足用户的需要，根据本地搜索的特点，可以更好地利用用户的搜索特征。文中提出通过对用户的行为分析，提取用户行为特征值，再运用排序学习的SVM（支持向量机）方法将分析得到的用户行为特征值融入本地搜索算法当中，以此实现对排序算法的优化。融人了用户行为特征后，本地搜索的排序结果平均准确率和前十名文档的相关性都有了一定的提高。实验结果显示，用户行为特征使得排序结果可以更容易、准确地反映用户的兴趣，提升了用户的搜索体验。

Optimizing of Local Search Ranking Algorithm Based on User Behaviors Analysis

With the development of the local search. the rank results of general ranking algorithm cannot fully meet the needs of users. The characteristics of local search make the possibility that the user＇ s search characteristics can be used more properly. By analyzing the user behaviors. the user behavior characteristic values are got. Then the SVM （Support Vector Machine） is employed to merge the user＇ s behavior characteristic values into the local search algorithm. And the ranking algorithm is optimized. The average accuracy rate of the local search rank results and the top ten documents correlation have improved to some extent, after integrated into the user behavior features. The experimental results show that user behavior features allow ranking results can more easily and accurately response the user＇ s interesting , and improve the user＇ s search experience.