基于语义扩展的搜索引擎研究

发布时间：2018-06-24 02:16

本文选题：语义搜索引擎 + 查询推理　；参考：《华东师范大学》2012年硕士论文

【摘要】：随着网络的快速发展,为了满足用户的信息需求,搜索技术和搜索引擎如雨后春笋,得到迅速发展。然而传统的搜索引擎返回的搜索结果往往不尽人意,在查全率和查准率上离用户需求差距很大。用户需要耗费大量时间去再次筛选搜索引擎返回的结果来寻找自己感兴趣的内容。随着语义网技术的发展,语义信息的积累,基于语义扩展的搜索引擎能够充分利用语义知识,对用户的查询进行语义扩展推理,并对搜索引擎的返回的查询结果进行语义文本自动摘要。更加准确的为用户提供有用信息的同时,还降低了用户的工作量,节约用户的时间。因此基于语义扩展的搜索引擎的研究对促进搜索引擎的发展,推进互联网技术有着重要的意义。论文在已有的语义搜索引擎基础上,提出自己完整的面向语义搜索的框架以及扩展。这个框架主要由查询预处理、推理机、本体服务器、本体查询扩展、信息采集与处理、索引器与索引数据库、自动摘要、结果排序与显示这八部分组成。在用户搜索关键词扩展方面,提出自己的算法,利用本体将用户搜索词构造出语义树,然后在语义树上计算语义距离,对搜索词进行合理扩展,它解决了查询词扩展规模过大,噪声数据过多的问题。在自动语义摘要方面改进了机械的关键词提取算法,针对他们现在存在的问题,提出了基于语义关系的摘要提取方法,首先将文本文档进行语义关系处理,构建语义层次结构图,计算出层次结构图中的概念权值,并筛选出关键概念。从而得到句子的重要度,确定摘要句,最终形成语义摘要。基于语义关系的文本摘要是对传统的机械统计摘要方法的改进,该方法具有较高的准确率很召回率在搜索结果排序方面,将传统的效果不理想的位置排序法引入了摘要排序法,在实际应用中取得了良好的效果。该方法在语义理解的基础上,整合了位置排序法,能够更加准确,合理有效的进行页面排序。
[Abstract]:With the rapid development of the network, in order to meet the information needs of users, search technology and search engines are springing up and developing rapidly. However, the search results returned by traditional search engines are often unsatisfactory, and there is a big gap between the recall rate and the recall rate. It takes a lot of time for users to re-filter the results returned by search engines to find the content they are interested in. With the development of semantic Web technology and the accumulation of semantic information, the search engine based on semantic extension can make full use of semantic knowledge to infer the query of users. And the query results returned by the search engine are automatically abstracted from the semantic text. More accurate to provide useful information to users, but also reduce the workload of users, save the time of users. Therefore, the research of search engine based on semantic extension is of great significance to promote the development of search engine and promote Internet technology. Based on the existing semantic search engine, this paper proposes a complete framework and extension for semantic search. The framework consists of eight parts: query preprocessing, inference engine, ontology server, ontology query extension, information collection and processing, indexer and index database, automatic summary, result sort and display. In the aspect of user search keyword extension, this paper puts forward its own algorithm, uses ontology to construct semantic tree of user search term, then calculates semantic distance in semantic tree, and extends the search term reasonably, which solves the problem that the scale of query word expansion is too large. The problem of excessive noise data. In the aspect of automatic semantic abstracting, the mechanical keyword extraction algorithm is improved. In view of their existing problems, a method of extracting abstracts based on semantic relation is proposed. Firstly, the text document is processed with semantic relation. Construct the semantic hierarchy diagram, calculate the concept weight in the hierarchy chart, and filter out the key concepts. In order to get the importance of the sentence, determine the summary sentence, and finally form the semantic summary. The semantic relation based text summary is an improvement on the traditional mechanical statistical summary method, which has a high accuracy and recall rate in the search result ranking. This paper introduces the traditional position sorting method, which is not ideal in effect, into the summary sorting method, and obtains good results in practical application. On the basis of semantic understanding, this method integrates the location sorting method, which can be more accurate, reasonable and effective.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.3

【参考文献】