面向期刊论文的搜索技术的研究与设计

Research and design of search technology for paper library

为了解决期刊论文的大量存储、PDF文本解析、无法快速准确搜索等问题,文章首先使用HDFS分布式系统存储大量期刊论文,其次采用PDF抽取技术将非纯文本格式的PDF期刊转换为纯文本格式,然后分析比较主流中文分词器,最后采用倒排索引、分布式索引和优化相关度排序算法等,分析并设计了一个面向期刊论文的搜索引擎。实验结果表明,本文有效的解决大量论文的存储和访问,缩短建立索引时间,提高了搜索结果的准确性。

In order to solve the problem of mass journal articles storage,PDF text parse,it is not fast and accurate search and other issues,at first this paper uses HDFS distributed system to store a large number of journal articles,secondly it uses PDF extraction technology to convert non-plain text format PDF journals into plain text format,and then it analyzes the mainstream Chinese word segmentation,and finally it uses the inverted index,distributed index and optimization relevance sorting algorithm,it analyzes and designs a search engine for journal papers. The experimental results show that it can effectively solve the storage and access of a large number of papers,shorten the indexing time and improve the accuracy of search results.