共查询到17条相似文献，搜索用时 539 毫秒
1.
2.
基于网络结构挖掘算法的引文网络研究 总被引：1，自引：0，他引：1
本文在对网络结构挖掘的两种典型算法(HITS算法和PageRank算法)进行比较分析的基础上,将PageRank算法应用到大规模引文网络中.对由236 517篇SCI文章构成的引文网络,计算得到每一篇文献的PageRank值,并深入分析了文献的PageRank值与通常使用的引文数指标之间的关系.分析表明:PageRank值具有与引文数很强的相关性和相似的幂律分布特征,但是PageRank算法能够在高引文文献中更好的区别文献的潜在重要性,并在很大程度上削弱作者自引对文献评价客观性的影响. 相似文献
3.
4.
陈定权 《现代图书情报技术》2004,20(2):42-45
首先分析了查找相关网页的一些相关算法，然后在标准的HITS算法基础上，提出了基于修正的HITS的查找相关网页算法。最后，通过实验来分析了这些算法的各自特点和不足。对基于超链分析的查找相关网页算法的研究，可以为用户提供一种新的检索和获取的信息的途径。 相似文献
5.
"链接工厂"欺骗(Link farm spam)和重复链接严重地损坏了像HITS这样基于链接排序的算法性能.为了检测与抑制Web"链接工厂"欺骗和重复链接,本文充分利用页面之间的复制信息,尤其是利用完全超链接信息识别可疑链接目标.提出一种由页面文档和完全链接构成的二部图结构,通过对二部图进行构建与分析,搜寻共享锚文本和链接目标的Web页面,在此过程中标识链接工厂和重复链接,并通过带惩罚因子的权重邻接矩阵减小可疑链接的影响.实时实验和用户仿真测试结果显示,本文算法能显著改善传统HITS类方法的信息搜索质量. 相似文献
6.
本文以PageRank算法和HITS算法为例,对基于超链接分析技术的搜索引擎排序算法进行分析,并总结了超链接分析技术应用于搜索引擎结果排序的局限性。 相似文献
7.
以PageRank算法和HITS算法为例,分析超链接分析技术的搜索引擎排序算法,总结超链接分析技术应用于搜索引擎结果排序的局限性。 相似文献
8.
丁一 《现代图书情报技术》2005,21(6):26-29
Web 信息检索（Information Retrieval）技术研究是应用文本检索研究的成果，它结合Web图论的思想，研究Web上的信息检索，是行之有效的Web知识发现的途径。传统HITS方法所获得的信息精确度相当低，而PageRank作为一通用的搜索方法，不能够应用于特定主题的信息获取。在充分分析了PageRank、HITS等现有算法和Web文档的相似度计算方法的基础上，提出了Web上查询特定主题相关信息发现的RG-HITS算法。它结合了Web超链接、网页知识表示的信息相关度以及HITS方法来搜索Web上特定主题的相关知识。 相似文献
9.
针对传统微博网络用户影响力度量方法中用户关系不明确、时间特征不明显的问题,从网络结构、交互行为和时间3个角度考虑影响力的个体差异,提出一种基于改进HITS算法的微博用户影响力计算方法。该方法以用户间动态交互行为作为切入点,选择其动态交互强度作为改进算法的边权重,通过sigmoid函数对中心值进行修正,整合用户的权威值和中心值量化用户影响力,提高算法的适应性,计算结果更为合理,为微博用户影响力计算提供了新的思路。以新浪微博为实验对象,实验结果表明,该方法相对于传统影响力计算方法,在DCG值、查准率、召回率和F值评价指标方面均有所提升。 相似文献
10.
权威因子:一个新的期刊评价指标 总被引：4，自引：0，他引：4
2009年版的中国科技期刊引证报告(核心版)新提出了一个权威因子指标,权威因子的计算采用的是PrestigeRank算法,PrestigeRank算法是在著名搜索引擎算法PageRank算法的基础上修改优化而成的,PrestigeRank算法专门针对期刊引用网络缺失严重提出了合理的解决方案,在权威因子的计算中,期刊之间的引用不再是同等对待,重要期刊的一次引用比次要期刊的一次引用要被赋予更高的权重.利用PrestigeRank算法计算2008年的中国科技论文与引文数据库(CSTPCD)的所有期刊的权威因子,并对权威因子与总被引频次进行了比较研究,发现总被引频次在一定程度上是流行性测度,而权威因子在更大程度上是权威性测度.我们也发现权威因子能部分消除因为学科规模和性质不同造成的被引机会的不同的影响. 相似文献
11.
The purpose of this study is to test for the presence of order-effect bias in journal ranking surveys. Data were obtained from 379 active knowledge management and intellectual capital researchers who rated 25 journals on a 7-point scale. Five different versions of the survey instrument were utilized. Consistent with the cognitive elaboration model, the satisficing theory, and the Gricean maxim of orderliness, order-effect bias was observed in journal ranking surveys. Journals that appear in the beginning of the ranking list delivered to survey respondents consistently receive higher scores than journals at the end of the list. Overall, the position of the journal in the list explains over 10% of its score. Therefore, authors of journal ranking studies are recommended to use multiple versions of the survey instrument with randomized journal orders. 相似文献
12.
以《复印报刊资料》(下称资料)研究论文定量分析和2000—2003年图书馆学情报学期刊全文转载排名为基础，通过比较其综合排名与《中文核心期刊要目总览》、《中文社会科学引文索引》、《中国人文社会科学核心期刊要览》的差异和CNKI、《中文科技期刊引文数据库》收录期刊基金论文和被引频次验证排名合理性，结果表明《资料》期刊排名有一定的合理性，基金论文和被引频次是衡量期刊排名的客观指标。 相似文献
13.
《Journal of Informetrics》2020,14(1):101007
This study compares the two-year impact factor (JIF2), JIF2 without journal self-citation (JIF2_noJSC), five-year impact factor (JIF5), eigenfactor score and article influence score (AIS) and investigates their relative changes with time. JIF2 increased faster than JIF5 overall. The relative change between JIF2 and JIF_noJSC shows that the control of JCR over journal self-citation is effective to some extent. JIF5 is more discriminative than JIF2. The correlation between JIF5 and AIS is stronger than that between JIF5 and the eigenfactor score. The relative change in journal rank according to different indicators varies with the ratio of the indicators and can be up to 60 % of the number of journals in a subject category. There is subject category discrepancy in the average AIS and its change over time. Through the screening of journals according to variations in the ratio of JIF2 to JIF5 for journals in individual subject categories, we found that journals in the same subject categories can have considerably different citation patterns. To provide a fair comparison of journals in individual subject categories, we argue that it is better to replace JIF2 with the ready-made JIF5 when ranking journals. 相似文献
14.
自然科学期刊自引对影响因子的"调控" 总被引：14，自引：0，他引：14
本文利用《中国科技期刊引证报告》,重新计算了其中几个学科的一些期刊除去自引后的影响因子,并对去除前和去除后的影响因子与期刊排名进行了对比,以考察期刊自引对影响因子和期刊排名的影响。调查发现目前个别期刊过度自引已经使期刊排名发生了失真。最后对如何遏制这种现象提出了一些建议。 相似文献
15.
Jerome K. Vanclay 《Journal of Informetrics》2012,6(1):19-26
Publication patterns of 79 forest scientists awarded major international forestry prizes during 1990-2010 were compared with the journal classification and ranking promoted as part of the ‘Excellence in Research for Australia’ (ERA) by the Australian Research Council. The data revealed that these scientists exhibited an elite publication performance during the decade before and two decades following their first major award. An analysis of their 1703 articles in 431 journals revealed substantial differences between the journal choices of these elite scientists and the ERA classification and ranking of journals. Implications from these findings are that additional cross-classifications should be added for many journals, and there should be an adjustment to the ranking of several journals relevant to the ERA Field of Research classified as 0705 Forestry Sciences. 相似文献
16.
This paper builds an index family, named bi-directional h-index, to measure node centrality in weighted directed networks. Bi-directional h-index takes the directed degree centrality as the initial value and iteratively uses more network information to update the node’s importance. We prove the convergence of the iterative process after finite iterations and introduce an asynchronous updating process that provides a decentralized, local method to calculate the bi-directional h-index in large-scale networks and dynamic networks. The theoretical analysis manifests that the bi-directional h-index is feasible and significant for establishing a greater conceptual framework that includes some existing index concepts, such as lobby index, node’s h-index, c-index and iterative c-index. An example using journal citation networks indicates that the bi-directional h-index is different from directed degree centrality, directed node strength, directed h-degree and the HITS algorithm in ranking node importance. It is irreplaceable and can reflect these measures of node’s importance. 相似文献