联系我们

论文查询结果	返回搜索

论文编号：	9878
作者编号：	1120110796
上传时间：	2017/12/22 16:35:43
中文题目：	基于本体的元搜索信息融合研究
英文题目：	The study in Meta-search Information Fusion Based on Ontologies
指导老师：	王知津
中文关键字：	信息检索；元搜索；本体；查询融合；结果融合
英文关键字：	information retrieval； meta-search；ontology；query fusion； result fusion
中文摘要：	在研究元搜索的过程中，本研究认识到元搜索信息融合的三个领域：数据集选择融合、查询融合与结果融合。由于本体已经有了较为成熟有效的理论研究，因此本研究主要是以本体为中心，以在数据集选择融合、查询融合与结果融合三个领域提出新方法为目的。通过对基于本体的元搜索信息融合进行理论梳理、体系构建以及方法评价，元搜索引擎能够在流程与性能方面得到很大提升，也使得数据集融合、查询融合与结果融合相互借鉴、对接，提高了研究结论的可靠性和实用性。本研究以网络环境下知识组织系统为研究范畴，以Web元搜索引擎技术为具体研究对象，将信息融合与本体技术结合起来，具体研究应用于表层网络和深层网络中的元搜索引擎的一些关键技术。主要的研究工作包括以下几部分。第一，分析了网络环境下知识组织系统的特点，强调语义学是情报学的重要理论基础。进而对元搜索理论、信息融合理论和本体理论的国内外研究现状进行了统计分析，阐释了各自的概念原理、方法技术及应用情况。第二，在本体谱的基础上，提出了本体在信息融合中的三种应用模式：基于本体的元搜索数据集选择模式、基于本体的元搜索查询融合模式以及基于本体的元搜索结果融合模式，梳理了本体与元搜索、信息融合之间的关联。第三，总结了查询融合基础，梳理了典型的元搜索查询融合算法：INQUERY方法、D-S证据理论，提出了基于本体的元搜索查询融合方案，对查询的不确定性进行语言建模，并进行实验分析。第四，总结了结果融合基础，梳理了典型的结果融合算法：倒数方法、Borda记数方法与Condorcet方法，在本体视角提出了基于可变N反馈的元搜索结果融合策略和算法，设计实验分析该算法的优劣。第五，构建了基于本体和信息融合的元搜索体系，并对基于融合的本体、基于本体的融合系统开发两个用例进行分析与评价。针对自然语言信息的加入，研究了基于本体的软信息融合方法。本研究的主要研究结论为：第一，本体与元搜索、信息融合之间存在较强的关联和相互作用，本体在信息融合中存在三种应用模式：基于本体的元搜索数据集选择模式，基于本体的元搜索查询融合模式，基于本体的元搜索结果融合模式。第二，基于本体的元搜索查询融合策略存在优势，能够在实验的对比分析中比现有融合算法在鲁棒性和查准率上有所改进。第三，基于本体的元搜索结果融合策略存在优势，能够在实验的对比分析中比现有融合算法在鲁棒性和查准率上有所改进。第四，本体不仅可以用于元搜索体系内部的融合，而且用在自然语言消息与元搜索体系融合上也是可以的。本研究的创新点主要体现为：第一，提出了基于本体的元搜索查询融合方案和算法，对查询的不确定性进行语言建模，并进行仿真实验分析，证明该方法在性能上比典型的查询融合方法有优势和改进。第二，在本体视角提出了基于可变N反馈的元搜索结果融合策略和算法，计算相关度系数，并在Web环境进行实验分析，证明该方法在性能上比典型的结果融合方法有优势和改进。第三，提出了基于本体和信息融合的元搜索体系，构建了基于本体谱的信息融合模型，并对基于融合的本体、基于本体的融合系统开发两个用例进行数据分析与评价。在体系内融合的基础上，引入自然语言信息与系统的融合，提出了基于本体的软信息融合理论，利用命题图将自然语言数据与已有的本体等多源数据相融合。
英文摘要：	In the study process of meta-search, the study recognizes three fields of meta-search information fusion: data set selection fusion, query fusion and result fusion. As the ontology has been a mature and effective theoretical research, the study is mainly focused on ontology application in meta-search fusion, expecting to propose new methods in the areas of data set selection fusion, query fusion and result fusion. Through the process of combing theories, constructing system and evaluating methods of meta-search information fusion based on ontology, meta-search engine can improve greatly in the processes and performances. It also reflects and connects among data set selection fusion, query fusion and result fusion, optimizing the reliability and practicability of the research conclusions. The study applies the knowledge organization system under the network environment as the research category, Web meta-search engine technology as the specific research object, combining information fusion with ontology technology. Some key technologies of meta-search engine applied in the surface Web and deepWeb are explored specificly. The main research work includes the following parts: First of all, the characteristics of knowledge organization system in the network environment are analyzed, and it is emphasized that semantics is the important theoretical foundation of information science. It analyzes statistically on the domestic and foreign research status of meta-search theory, information fusion theory and ontology theory, and defines their concepts, principles, methods, technologies and application situation. Secondly, the relationships between ontology and meta-search and information fusion is combed. On the basis of the spectrum of ontologies, it puts forward three application modes of ontologies application in information fusion: meta-search data set selection mode based on ontology, meta-search query fusion based on ontology, and meta-search result fusion mode based on ontology. Thirdly, it summarizes the basic theory of meta-search query fusion, combs the typical fusion algorithms: INQUERY method, D - S evidence theory, and puts forward the meta search query fusion scheme based on ontology along with language modeling and experimental analysis . Fourthly, it summarizes the basic theory of meta-search result fusion, combs the typical fusion algorithms: reciprocal method, Borda notation method and Condorcet method. In the ontology perspective, the fusion strategy and algorithm of meta-search result fusion based on variable N feedback is puts forward, along with experiment analysis and improvement on robustness and average precision. Fifthly, a meta-search system based on ontology and information fusion is constructed, and two use cases of ontology based fusion and fusion system on ontology are analyzed and evaluated. Towards the introduction of natural language data, it proposes soft information fusion theory based on ontology. The main conclusions of the study are: First, there is a strong correlation and interaction between ontology and meta-search and information fusion. Ontology has three application modes in meta-search fusion. Second, meta-search query fusion scheme and algorithm based on ontology carries on experimental analysis in the Web environment, proved that more robustness and average precison than any typical query fusion method. Third, meta-search result fusion scheme and algorithm based on ontology carries on experimental analysis in the Web environment, proved that more robustness and average precison than any typical result fusion method. Fourth, not only applies ontology in the system of meta-search fusion, but also applies in the fusion of natural language message and meta-search system. The innovation points of the study are mainly reflected in: Firstly, a meta-search query fusion strategy and a concrete fusion algorithm based on ontology is proposed. It carries out language modeling towards query uncertainty, and simulation experiment proved better performance than typical methods. Secondly, in the ontology perspective, the fusion strategy and algorithm of meta-search result fusion based on variable N feedback is puts forward, computs the correlation coefficient, along with experiment analysis and improvement on robustness and average precision in the Web environment proved better performance than any typical result fusion method. Third, a meta-search system based on ontology and information fusion is constructed, and two use cases of ontology based fusion and fusion system on ontology are analyzed and evaluated. In the base of meta-search system fusion, towards the introduction of natural language data, it proposes soft information fusion theory based on ontology, using propositional graph to integrate natural language data with existing ontology of multi-source data.
查看全文：	预览  下载(下载需要进行登录)