【作者】 贾鹏；

【导师】 王祎；

【作者基本信息】 大连理工大学 ， 软件工程（专业学位）， 2012， 硕士

【摘要】 科技文献是科研工作者用来发布研究成果、交流信息的重要载体。而科技的迅速发展和互联网的广泛普及,使得电子文献海量涌现,所以急需一个能够辅助科技工作者快速有效获取相关文献,并对其进行分析和管理的文献检索系统。本文运用网络信息抽取、自然语言处理、机器学习等领域的理论和方法,结合启发式规则,构建了一个英文PDF科研文献的检索系统。主要工作集中在以下几个方面：文献信息的网络抽取、文献元数据抽取、主题发现与分析。在网络信息抽取方面,本文利用网络数据抽取工具Web-Harvest,收集指定页面信息。根据相应规则,得到科技文献的相关信息。在文献元数据抽取方面,本文采取机器学习与规则相结合的思路提取科技文献元数据信息,从而尽量保证了算法的鲁棒性和精度。在元数据准确提取的基础上,本文通过主题模型,发现文献的主题；并在在时间域上分析主题,得到领域内特定主题的发展历程。总之,本文实现的科技文献分析与检索系统,为科技工作者提供文献检索、内容分析、可视化等帮助,系统本身具有较高的实用价值。更多还原

【Abstract】 The scientific and technical literature becomes main method for researchers to learn the state-of-the-art academic achievements and publish their research results. However, this development also results in massive data. So urgently we need a literature retrieval system that can auxiliary researchers’fast effective access to relevant literature, and carries on the analysis and management of the literature retrieval system.This paper uses theories and methodologies related with Web information, natural language processing and machine learning, combined with heuristic rules, and construct an English PDF research literature retrieval system. This paper focus on the following several aspects:web information extraction, metadata extraction of scientific literature, topic discovery and topic analysis.In terms of web information extraction, we use network data extraction tools Web-Harvest, collecting information on a given page. According to relevant rules, get information about scientific literature. About metadata extraction of scientific literature, by combining the machine learning and heuristic rules, we get the metadata quickly and accurately.Accurately extract the metadata, based on the theme of this paper model, we found that the subject of literature; and analysis of themes in the time domain to obtain a specific topic within the field of the development process.In a word, this paper realizes an analysis and retrieval system of scientific literature, which is efficient and friendly with interface, it would be a useful assistant for researchers in their studies.更多还原