基于面本体的情境搜索研究

发布时间：2018-05-29 16:03

本文选题：面本体 + 情境本体　；参考：《西南大学》2013年硕士论文

【摘要】：人们在传统搜索引擎上所用到的关键字搜索技术发展成熟,是当今最流行的搜索方式。但是用户输入关键字,搜索自己想要的东西,想表达的意思仅仅用几个关键词并不能完全表达清楚,而计算机却只能根据关键词作全文匹配,常常不能达到用户想要的效果。比如,“椅子”这个词,它本身的属性包括了椅子的材质,形状,设计工艺,制作厂家,制作时间,地点等等,这一系列潜在的属性构成了一个复杂的基于语义的关联网。单纯的关键词匹配已经不能适应人们对搜索更精细化,人性化的要求。下一代网络WEB3.0”语义网”在搜索方面的研究侧重于具有语义分析能力的搜索引擎,要求搜索引擎能够理解人类的自然语言,并且具有一定的推理和判断能力。本文提出了一种基于面本体的情境搜索模型(Context Search Based on Faceted Ontology, CSFO), CSFO模型构建面本体FO来表示信息搜索领域的各个属性面,构建情境本体CO来创造用户搜索的当前情境,CSFO模型的主要思想根据情境本体CO对用户输入的关键字进行语义扩展,生成相应的搜索本体SO,通过面本体FO标注独立数据源的异构文档,最后利用相关的本体匹配算法进行搜索本体与面本体的本体匹配,从面本体中匹配出与当前搜索本体最相似的结果本体RO,实现情境搜索。CSFO模型旨在为用户提供更符合用户当前情境的搜索结果。本文的研究工作主要包括以下五个方面：第一,本文将本体技术整合到信息检索中,构建多个不同的本体数据模型表示不同的信息。本文用到的本体数据模型包括：情境本体CO,搜索本体SO,面本体FO,结果本体RO。CO用来表示用户的情境信息,SO来表示用户当次的查询信息,FO来表示被搜索的相关领域信息,RO用来表示当次的查询结果。第二,本文用搜索本体模型SO代替关键词进行搜索查询,将用户情境信息引入搜索系统。本文利用情境本体模型中的概念词汇,对用户在客户端输入的关键词进行语义扩展,规范用户自然语言的查询输入,提取有检索意义的概念词组通过基于情境本体的语义扩展算法SECO生成完成当次搜索的搜索本体模型。用搜索本体代替关键词进行信息匹配,扩大了用户情境信息表示的完整度。第三,本文从不同的独立数据源构建面本体FO,规范搜索信息的表示形式。本文通过下面四个步骤：a.产品系列分类；b.实体扩展；c.生成面单元FU及概念映射；d..面建模；来构建一个完整的特定领域的面本体。每个面本体从实体(Entity),性质(Property),活动(Action),修正(Modifier)这四个面进行概念分层。本文将被搜索的信息用面本体进行表示,不仅能规范各个独立数据源的信息,而且能通过面本体扩展各实体,概念及其属性之间的相关关系,丰富情境搜索结果的内容,提高搜索结果与用户情境的相关性。第四,本文通过搜索本体SO与面本体FO的本体匹配从用户角度,返回查询信息,实现情境搜索。本文用到两种本体匹配的方法进行搜索本体与面本体的本体匹配,其中一种是基于图的情境搜索算法CRG,该算法借鉴了相似度洪泛算法的思想；另一种是不一致本体推理算法CRIO,是将面本体,情境本体,搜索本体,结果本体全部抽象成不一致本体,通过不一致本体的推理,从面本体中匹配出与之相似的搜索本体。无论是哪一种方法,最终目的都是在面本体中匹配出与搜索本体最相似的面本体的子本体,然后通过相关函数将子本体转换成由标题,日期,链接,内容这几个关键字段组成的数据信息,形成搜索结果,返回给用户实现情境搜索。第五,实验分析与评价。为验证CSFO模型的可行性和有效性,通过仿真实验对该方法进行了小数据集的模拟与实现,同时,通过即有的平台数据集进行对比分析,实验结果表明,本文提出的方法应用于搜索引擎,与传统的信息搜索比较,具有更好的查全率与查准率。本文的研究对语义搜索具有一定的促进作用,对推进语义搜索的应用研究从理论层面过渡到工业实现层面也有积极的意义。
[Abstract]:The keyword search technology used by people in the traditional search engine is mature and is the most popular search method in the present day . However , the user inputs the keyword , searches for what you want , and the computer can only use several keywords to fully express clearly , but the computer can only do the full - text match according to the keyword , and the computer can not meet the requirements of the user . For example , the research of the next generation network WEIB3.0 " Semantic Web " focuses on the search engine with semantic analysis capability , which requires the search engine to understand the natural language of the human beings , and has certain reasoning and judgment ability .

A kind of context search based on Faceted Ontology ( CSFO ) is presented in this paper . The main idea of CSFO is to create user ' s current situation . The main idea of CSFO model is to create user ' s current situation . The main idea of CSFO model is to make semantic extension of key words input by user based on context ontology CO .

Firstly , the ontology technology is integrated into the information retrieval to construct a plurality of different ontology data models to represent different information . The ontology data model used in this paper includes : context ontology CO , search ontology SO , face ontology FO , result ontology RO . CO is used to represent user context information , SO to represent user context information , FO to represent the relevant field information searched , and RO is used to represent the results of the query .

secondly , the search query is carried out by using the search ontology model SO in place of the keyword , and the user context information is introduced into the search system .

Third , this paper constructs surface ontology FO from different independent data sources , canonical search information representation form . This paper uses the following four steps : a . product family classification ;
b . Entity extension ;
c . generating surface unit FU and concept mapping ;
d. Surface modeling ;
In this paper , the information of each independent data source can be regulated , and the relationship between each entity , concept and its attributes can be extended through the surface ontology , and the content of the results of the situation search can be enriched , and the relevance of the search result and the user context can be improved .

Fourthly , this paper uses two kinds of ontology matching methods to match the ontology to the ontology , one of which is the context search algorithm CRG based on the graph , which uses the idea of similarity flood algorithm .
The invention relates to an inconsistent ontology reasoning algorithm CRIO , which is characterized in that a face body , a situational ontology and a searching body are all abstracted into inconsistent bodies ;

Fifth , experimental analysis and evaluation . In order to verify the feasibility and effectiveness of CSFO model , the simulation and realization of this method are carried out by simulation experiment .

The research of this paper has a certain promotion effect on the semantic search , and the application research on the promotion of the semantic search has a positive meaning from the theoretical level to the industrial realization level .
【学位授予单位】：西南大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.1

【参考文献】