【摘要】 为了提高万维网信息检索的性能,该文提出在基于机器学习的检索排序算法中,利用万维网链接图的拓扑特性来度量网页的重要性。链接图的拓扑特性是从各种不同的角度对链接图结构信息的描述,利用它来度量网页的重要性可以更加全面地利用链接图所包含的结构信息,并且能够更好地发挥学习机的学习能力。通过实验表明,在基于排序提升器的检索排序算法中,利用链接图的拓扑特性可以非常有效地度量网页的重要性,其性能要优于传统的PageR ank算法。更多还原

【Abstract】 In order to improve the performance of web information retrieval,this paper proposes a method for measuring the importance of web page from the topological properties of web graph in the learning framework.The topological properties describe the web graph from a variety of viewpoints so the analysis takes full advantage of the structural information contained web graph and the learning ability of the learning machine.Experimental results indicate that the method effectively measures the importance of Web pages from the topological properties of the Web graph,and gives better performance than the PageRank algorithm.更多还原