【作者】 王玮；

【导师】 徐秀娟；

【作者基本信息】 大连理工大学 ， 软件工程， 2016， 硕士

【摘要】 近年来,由于智能交通和人们的生活以及社会的经济息息相关,所以这个领域涌现出大量的文章,是最活跃的研究领域之一。尽管有许多学者和机构进行过文献研究,但只运用了基础的统计和简单的算法进行分析。因此,很有必要对该领域进行系统深入地分析。本文提出一个多角度文献分析框架ITS-Frame,从多个方面对智能交通领域的文献展开分析。(1)基础文献分析：利用统计学的方法,找到发表文章数目最多的作者、机构和国家。(2)影响力分析：文章影响力方面,从文章的引用数和NCII指数两方面进行分析；作者影响力方面,用APS、ACS两个指数进行排序,并提出基于PageRank的AuthorRank算法,比较了作者的权威度。(3)合作模式：在传统GN算法基础上,为了平衡边介数和权重之间的关系,提出基于边值的合作网络分析算法,分别对作者合著网络、同现关键词网络、作者间共关键词网络进行构造和分析。此外,还在时间序列上分析作者合著关系的变化,进而挖掘作者研究方向的变化。(4)话题分析：本文提出了LK-means主题模型,使用LDA进行文本降维,通过衡量文档间的差异,对文章主题及其变化趋势进行挖掘。本文使用爬虫技术从IEEE和ELSEVIER网站上,抓取了5个智能交通领域的数据集。通过多角度地分析,发现了该领域科学家的合著网络变化,也挖掘了一些活跃地话题,如GPS、交通控制和道路安全等。除了美国,还发现中国、欧洲和一些汽车制造企业对这个领域的贡献也越来越大。总之,本文的工作为该领域的研究者提供了一个全面的文献分析。更多还原

【Abstract】 Since there is a close relationship between intelligent transportation systems and our daily life, there are a large number of related papers appear in this field. And it is one of the most active research fields in recent years. Although many scholars and institutions carried out bibliographic analysis, they only based on simple statistical methods and data mining algorithms to study some aspects they want to know. Therefore, it is necessary to give an in-depth analysis about this field.Our paper presents a framework named ITS-Frame for multi-aspect analysis. (1) Basic bibliographic Analysis. We identify the most productive authors, institutions and countries in the publications of ITS. (2) Impact analysis. In the aspect of paper influence, we rank the papers from citations and NCII index. In the aspect of author influence, we rank authors based on APS, ACS index, and present an AuthorRank algorithm based on PageRank to compare authors’authority. (3) Collaboration Pattern analysis. Based on traditional GN algorithm, we present a collaboration pattern analysis algorithm which can balance edge betweenness and weight. And we use it to construct and analyze three networks, including co-authorship network, keyword co-occurrence network and author co-keyword network. In addition, we analyze the co-authorship network in a time series, and know the change of experts’research direction. (4) Topic analysis. We present a novel algorithm based on LK-means to find the topic of every paper and show the evolution of themes. We use LDA to reduce text dimension and measure the difference between documents.We obtain five experiment dataset in the field of ITS from IEEE and ELSEVIER website using a web crawler. By multi-aspect analysis, we find the changes of some experts’ co-authorship networks. We also identify some active keywords, such as GPS, traffic control, road safety. Besides USA, we find that China, European countries and some automobile manufacturers also play significant roles in this field. In conclusion, our work can provide a comprehensive analysis to other researches in ITS fields.更多还原