基于多核聚类算法和用户兴趣模型的图像搜索方法

Image Search Method based on Multi-kernel Clustering Algorithm and User Interest Model

现有的搜索引擎算法不能完整地分析用户的查询目的,直接影响了检索的质量并增加了用户检索的代价。为了提高图像搜索的效率,文中提出了一种基于多核聚类算法的图像搜索方法,通过使用最小二乘支持向量机建立用户兴趣模型,并将个性化的搜索结果返回给用户。实验证明,与其他聚类算法相比,该算法的查全率和查准率可分别提升了8.2%、11.42%和19.7%、26.08%,有明显的提升效果。

Traditional search engines can’t completely evaluate user’s search aim.It will lead to retrieval quality decline and increased cost.The paper proposed a new personalized image searching algorithm based on clustering analysis and user interest model.It utilizes relevant feedback and LSSVM to build user interest model and return the personalized searching results to the users based on the muli-kernel cluster for images.The analysis of experiment results indicates that compared with the traditional searching algorithm with single feature the improved algorithm can increase the mean recall and precision ratio obviously.