基于情景感知融合的个性化农业信息推荐服务研究

发布时间：2018-01-10 10:39

本文关键词：基于情景感知融合的个性化农业信息推荐服务研究　出处：《南京农业大学》2015年硕士论文　论文类型：学位论文

【摘要】：农业信息化是我国农业现代化发展的重要推动力量,伴随其不断发展的过程中产生的相关信息资源数量呈指数级别快速增长,造成现代信息技术水平和知识水平偏低的农业用户在寻找所需的信息资源时会遇到信息过载和迷向的问题,不利于我国农业信息化建设工作进一步开展。随着我国经济不断发展,农民生活水平日益提高,各类农业信息化基础设施建设进一步完善,利用现代化信息技术实现农业个性化信息资源推荐服务,为用户针对性提供及时、有效的信息资源成为一种行之有效的解决方案。本文针对现实生活中的实际需求以及农业信息服务体系中出现的问题进行探讨分析,提出了一套基于情景感知融合的个性化农业信息推荐服务的研究方案,解决在农业信息资源推荐服务过程中用户兴趣模型建立与更新、信息资源模型建立与分类以及个性化推荐服务这三个核心模块遇到的问题,提高了个性化信息推荐服务的性能与准确率。本文主要工作如下:(1)综述了农业信息资源领域关于基于情景融合的个性化推荐相关理论知识与算法,对于农业信息服务体系现状进行分析,从信息推荐服务的信息服务用户、信息服务内容以及信息服务渠道方式等环节进行分析,发现农业信息推荐服务存在的问题及需求,构建基于情景感知融合的个性化农业信息推荐服务框架。(2)利用基于分类因素和位置因素改进的TFIDF算法对农业信息资源进行特征提取,建立多维农业信息资源模型和农业信息资源特征词典,借助多维分类向量空间模型从类型、时空和对象三个维度描述信息资源不同维度的特征,完成对农业信息资源的获取、建模和分类。(3)依据农业领域知识建立的规则将感知到的用户情景信息融入到用户兴趣模型中,结合显式和隐式获取以及最近邻用户协同过滤推荐的兴趣信息,在用户类型维和对象维两个维度上完成对于用户兴趣模型的初始化和更新。(4)采用Java和MySQL技术将信息资源模型、用户兴趣模型以及基于内容的个性化推荐予以实现,为基于情景感知融合的个性化农业信息推荐服务框架提供解决设计方案。结合真实数据对本文提出的个性化信息服务与传统的农业信息服务进行对比,在体现用户兴趣迁移的敏感性、准确性以及推荐信息的精准率、召回率以及F-测度值三个评价指标均有提高。本文的创新工作主要由以下两个方面组成:第一、将用户的情景信息感知、关联、融合到用户兴趣模型当中,结合农业领域知识,提出了一种对于情景信息新的处理方法,发挥普适计算的优势,让计算机可以更加精确、实时的获取用户的个性化需求,提高农业信息资源推荐服务的精确度。第二、采用分类向量的方法对于农业信息资源以及用户兴趣模型进行表示,既能够体现出农业信息资源的周期性、地域性、动态性、实用性、科学性特点,又可以进一步细化区分特征项,描述农业信息资源在不同维度的特性,排除不同维度下相同特征词的相互干扰,提高模型表示精度以及推荐效率。
[Abstract]:Agricultural informatization is an important force to promote the development of agricultural modernization in China, with the number of information resources related to the production process of continuous development in exponential level rapid growth caused by modern information technology and knowledge level of the low level of agricultural users will encounter many problems of information overload and to find the needed information resources, is not conducive to the work of agricultural informatization in China. With the further development of China's economic development, the improvement of the living standard, the construction of agricultural information infrastructure to further improve the implementation of personalized information recommendation service of agricultural resources utilization of modern information technology, for the user to provide timely and effective information resources has become an effective solution in this paper. According to the analysis of real life needs and problems in the agricultural information service system, Put forward a set of personalized service recommendation scheme based on context aware agricultural information integration, solve in the Agricultural Information Resource Recommendation Service in the process of establishing and updating the user interest model, and the classification and personalized service is the three core module problems of establishing information resources model, improves the performance and accuracy of the personalized information recommendation service rate. The main work is as follows: (1) review of agricultural information resources on the recommendation of related theoretical knowledge and algorithm of scene fusion based personalized, the present situation of agricultural information service system of information service user recommendation service from the information content, link information service and information service channels and analysis, found the problem and demand the existence of the agricultural information service, construct a framework for personalized recommendation service of agricultural information fusion based on context awareness. (2) by using the improved TFIDF algorithm based on the classification of factors and location factors of agricultural information resources for feature extraction, multidimensional model and characteristics of agricultural information resources of agricultural information resource dictionary, using multidimensional classification vector space model from the type, characteristics of the three dimensions of time and space and object description of information resources in different dimensions, complete access to agricultural information resources the modeling and classification. (3) on the basis of agricultural knowledge rules established the user perceived contextual information into the user interest model, combined with the explicit and implicit user access and nearest neighbor collaborative filtering in information, in the two dimensions of user type dimension and objective dimension for user interest model the initialization and update. (4) using Java and MySQL technology to the information resource model, user interest model and content based personalized recommendation to be realized, which is based on The framework of context aware personalized recommendation service of agricultural information fusion design solutions. Based on the proposed real data of the personalized information service and the traditional agricultural information service were compared in sensitivity to reflect user interest transfer, information accuracy and recommendation accuracy rate, recall rate and F- value of three evaluation indicators have improved. The innovation of this paper is mainly composed of the following two parts: first, the scene information perception, user association, integrated into the user interest model, combined with agricultural knowledge, proposes a new method for processing context information, play the advantages of pervasive computing, let the computer can be more precise, personalized needs of users access in real time, improve the agricultural information resource service recommendation accuracy. Second, the agricultural information resources and using the method of vector classification The user interest model. Not only can reflect the periodicity of agricultural information resources in regional, dynamic, practical and scientific characteristics, and can further refine the distinguishing features, characteristics of agricultural information resources described in different dimensions, eliminate mutual interference of different dimensions of the same character, improve the model accuracy and the efficiency of recommendation.

【学位授予单位】：南京农业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F323.3;S126

【相似文献】