个性化信息获取方法的研究

发布时间：2018-07-01 14:19

本文选题：个性化 + 数据挖掘　；参考：《大连理工大学》2004年博士论文

【摘要】：随着Internet技术的发展,信息获取对于人们的工作生活越来越重要。目前,针对极度膨胀的信息资源,人们主要使用搜索引擎(Search engine)或是一些智能代理软件(Information retrieval system and filtering system or Agent)来获取网上的信息资源。但是由于传统查询模型的限制,信息搜索的精度不高是人们经常遇到的问题,因此使用“用户特征信息”进行个性化信息搜索是今后信息获取工具的发展方向。但是目前不论是个性化信息需求特征的研究、个性化信息挖掘算法的研究、还是个性化信息获取系统的研究都存在很大的不足。本文针对个性化信息获取的问题,从几个方面对其相应的理论与算法进行了研究,主要研究工作如下: (1) 个性化信息获取特点和方法的研究。首先研究了个性的概念,然后分析了个性化信息获取的特点和方法,探讨了在个性化信息获取过程中个性化知识的运用,提出了用户特征模板的概念,进行了模板基本结构及应用分析,并给出了模板的示例。 (2) 个性化信息获取实证方法的研究。对个性化信息获取系统的实验方法进行研究,充分考虑由于个性信息的差异带来的干扰因素,给出一个通用的、无实验者偏见的,对个性化信息获取系统和算法进行检验的实验方法。从查询结果的个性化评价角度,建立了对个性化信息服务系统性能及算法效率进行评价的指标体系。 (3) 基于奇异值分解的个性化信息挖掘算法的研究。本文在研究了数据挖掘和模式识别技术在个性信息挖掘中的应用之后,提出了基于奇异值分解的个性化信息挖掘算法,并应用此算法进行了个性化信息挖掘实验,分析及提出了算法改进的方向。 (4) 奇异值分解算法和神经元网络法相结合的模式识别算法的研究。研究了神经网络法,并结合SVD算法,提出一个使用用户信息需求特征构造个性化空间,同时改良样本空间和搜索空间,进行个性化信息检索的算法。 (5) 应用遗传算法进行个性特征提取算法的研究。个性化信息获取方法的研究研究了特征提取算法和遗传算法,提出了两种基于用户特征文档集合的用户个性特征提取算法,并通过实验验证了算法的效率。本文通过对个性化信息获取的评价方式、实验方法和挖掘算法等几方面的研究,为今后个性化信息服务打下了基础。关键词:个性化;数据挖掘;模式识别;信息获取;文本分类算法;奇异值分解;
[Abstract]:With the development of Internet technology, information acquisition is more and more important for people's work and life. At present, for the extremely inflated information resources, people mainly use search engine (search engine) or some intelligent agent software (Information retrieval system and filtering system or Agent) to obtain information resources on the Internet. However, because of the limitation of traditional query model, the low precision of information search is a common problem, so it is the development direction of information acquisition tool to use "user characteristic information" to carry out personalized information search. However, there are many shortcomings in the research of personalized information requirements, personalized information mining algorithms and personalized information acquisition systems. Aiming at the problem of obtaining personalized information, this paper studies its theory and algorithm from several aspects. The main research work is as follows: (1) the characteristics and methods of personalized information acquisition. This paper first studies the concept of personality, then analyzes the characteristics and methods of obtaining personalized information, discusses the application of personalized knowledge in the process of obtaining personalized information, and puts forward the concept of user feature template. The basic structure and application of template are analyzed, and an example of template is given. (2) the empirical method of personalized information acquisition is studied. The experimental method of personalized information acquisition system is studied, and the interference factors caused by the difference of personality information are fully taken into account, and a general and non-experimenter bias is given. The experimental method to test the personalized information acquisition system and algorithm. From the perspective of personalized evaluation of query results, An index system for evaluating the performance and algorithm efficiency of personalized information service system is established. (3) the research of personalized information mining algorithm based on singular value decomposition (SVD). After studying the application of data mining and pattern recognition technology in personality information mining, a personalized information mining algorithm based on singular value decomposition is proposed in this paper. The improvement direction of the algorithm is analyzed and proposed. (4) the research of pattern recognition algorithm based on the combination of singular value decomposition algorithm and neural network method. In this paper, the neural network method is studied and combined with the SVD algorithm, a personalized space based on user information requirement features is proposed, and the sample space and search space are improved simultaneously. (5) the application of genetic algorithm to personality feature extraction algorithm. The research of the personalized information acquisition method has studied the feature extraction algorithm and the genetic algorithm. In this paper, two algorithms for extracting user personality feature based on user feature document set are proposed, and the efficiency of the algorithm is verified by experiments. In this paper, the evaluation methods, experimental methods and mining algorithms of personalized information acquisition are studied in order to lay a foundation for personalized information service in the future. Keywords: personalization; data mining; pattern recognition; information acquisition; text classification algorithm; singular value decomposition;
【学位授予单位】：大连理工大学
【学位级别】：博士
【学位授予年份】：2004
【分类号】：F062.5

【引证文献】