基于文本相似度和入出比的改进PageRank科研文献搜索方法

Improved PageRank Search Method Based on Text Similarity and in-out Ratio

为针对性筛选出契合具体研究方向的高质量学科论文,在PageRank算法基础上,提出一种考虑文本相似度和节点入出比的CSTPR(Content similarity thresholded PageRank)文献搜索方法。构建学科文献的引用关系网络,并计算节点入出比及文本相似度,算法迭代得到CSTPR。以超润滑相关研究文献为例,实验分析验证了CSTPR的有效性,算法能快速获得内容契合度高、代表性强的优秀文献。

In order to select high-quality articles that match specific research needs,a search method called CSTPR(content similarity threshold PageRank)is proposedThis method takes text similarity and in-out ratio into consideration based on PageRank algorithmFirst,citation network of subject literature is constructedThen,the inout ratio of nodes and text similarity are calculated for screening threshold and weight matrixFinally,CSTPR of nodes is computed for literature ranking by iterative calculationThe experimental results show the effectiveness of the improved search methodCSTPR model can quickly find highly content-fitted and strongly representative literature.