共查询到20条相似文献，搜索用时 218 毫秒
1.
推荐系统是电子商务系统中最重要的技术之一.随着电子商务系统用户数目和商品数目日益增加,在整个商品空间上用户评分数据极端稀疏,传统的相似性度量方法均存在各自的弊端,导致推荐系统的推荐质量急剧下降.针对用户评分数据极端稀疏情况下传统相似性度量方法的不足,本文提出了一种基于相似项目与用户评分预测的协同过滤推荐算法,综合利用相似项目和相似用户评分信息预测用户对未评分项目的评分.通过聚类算法形成用户候选近邻集,减小了算法搜索空间,降低了最近邻用户的搜索时间,从而增强了算法的扩展性.实验结果表明,本算法可以有效解决用户评分数据极端稀疏情况下传统协同推荐算法存在的问题,显著提高推荐系统的推荐质量. 相似文献
2.
基于粗糙用户聚类的协同过滤推荐模型 总被引：1，自引：0，他引：1
【目的】将粗糙集引入到基于用户聚类的协同过滤中,提高推荐质量。【方法】提出一种基于粗糙用户聚类的协同过滤推荐模型:离线时采用粗糙K-means用户聚类算法,根据用户与聚类中心的相似度将其分配到K个类的上、下近似中,形成用户的初始近邻集;在线时从目标用户的初始近邻集中搜索其最近邻,预测项目评分并向其产生推荐。【结果】通过实验对比发现,该模型比传统的和基于项目的协同过滤推荐算法降低约14%的平均绝对误差,比基于用户聚类的协同过滤推荐算法降低约10%的平均误差。【局限】在考虑上、下近似对聚类中心调整的重要程度时,忽略了用户聚类数目和最近邻集用户数阈值的变化所产生的影响。【结论】该模型能有效提高推荐精度,具有较强的可行性和现实意义。 相似文献
3.
目前协同过滤被广泛应用于数字图书馆、电子商务等领域的个性化服务系统.最近邻算法则是最早提出和最主要的协同过滤推荐算法,但用户评分数据稀疏性严重影响推荐质量.针对上述问题,提出了一种基于Rough集理论的最近邻协同过滤算法,以用户评分项并集作为用户相似性计算基础,并将非目标用户区分为无推荐能力和有推荐能力两种类型;对于无推荐能力用户不再计算用户相似性以改善推荐实时性,对于有推荐能力用户则提出一种基于Rough集理论的评分预测方法来填补用户评分项并集中的缺失值,从而降低数据稀疏性.实验结果表明新算法能有效提高推荐质量. 相似文献
4.
5.
融合社会网络的协同过滤推荐算法研究 总被引：1，自引：0，他引：1
针对传统协同过滤推荐算法的数据稀疏性及恶意行为等问题,提出一种新的基于社会网络的协同过滤推荐算法。该算法借助社会网络信息,结合用户信任和用户兴趣,寻找目标用户最近邻居,并以此作为权重,形成项目推荐,以提高推荐的准确度。实验表明,相对于传统的协同过滤算法,该算法可有效缓解稀疏性及恶意行为带来的问题,显著提高推荐系统的推荐质量。 相似文献
6.
协同过滤技术是推荐系统中核心技术之一,数据的稀疏性和用户的多兴趣性困扰着协同过滤推荐质量的提高.将用户相似性和项目相似性结合起来,对原始评价矩阵进行降维处理,得到对目标评价预测影响最大、数据规模非常小的最近邻评价矩阵,在该矩阵上依照项目近邻程度不同对目标评价预测贡献不同的方法,对用户的邻居进行加权精选,对目标评价实现交错预测.实验结果验证该算法能达到较高的推荐精度. 相似文献
7.
基于项目分类预测的协同过滤推荐算法 总被引：3，自引：0，他引：3
在电子商务系统中,为了帮助用户有效地发现、过滤和利用信息,信息过滤技术应运而生.协同过滤技术作为其中的一种技术被成功地应用于推荐系统中.随着电子商务用户数目和商品数目的日益增加,整个项目空间上用户评分数据极端稀疏,传统的相似性度量方法均存在各自的弊端,导致推荐系统的推荐质量急剧下降.针对这一不足,提出基于项目分类预测的协同过滤算法,通过对用户评分矩阵中的项目进行相应的分类,缩小邻近搜索的范围,预测项目评分,减少稀疏性,并采用新的相似度计算方法.实验结果表明,该算法能提高个性化推荐算法的准确性. 相似文献
8.
基于属性值偏好矩阵的协同过滤推荐算法 总被引：7，自引：2，他引：5
传统的协同过滤推荐算法面临用户评分数据稀疏性和冷启动问题的挑战.针对上述问题,提出了基于属性值偏好矩阵的协同过滤推荐算法,首先采用奇异值分解(SVD)对用户-项目评分矩阵降维得到目标用户的初始邻居用户集,生成新的用户-项目评分矩阵;然后将用户评分映射到相应的项目属性值上,生成每个用户的属性值偏好矩阵,并基于属性值偏好矩阵进行用户相似性度量,从而缓解了评分数据稀疏性;将新项目的属性值与用户的属性值偏好矩阵进行匹配,从而找出匹配度最高的前N个用户作为新项目的推荐受众.实验结果表明了该算法的有效性. 相似文献
9.
10.
在电子商务中,协同推荐技术能够帮助用户发现感兴趣的东西.在协同推荐中,通常采用最近邻居的方法来产生推荐.随着商品数量的增多,协同推荐所需要的数据集也越来越稀疏,可用数据比例越来越少.为了解决这个问题,本文在传统的评分数据的基础上,引入用户的基本信息,对用户的基本信息进行离散化处理,将用户的基本信息转化成一个0、1的向量,在用户的信息的基础上计算最近邻居,根据最近邻居对用户缺失数据进行补充,在补充后的评分数据上进行聚类计算,并根据聚类结果对用户评分进行预测.实验表明引入用户的基本信息,并采用对基本信息离散化的处理方式进行缺失数据补充,在此基础上进行数据的聚类,能够提高预测评分的准确性. 相似文献
11.
12.
文献推荐系统：提高信息检索效率之途 总被引：2，自引：0，他引：2
Traditional Information Retrieval (IR) systems have limitations in improving search performance in today’s information environment. The high recall and poor precision of traditional IR systems are only as good as with the accuracy of search query, which is, however, usually difficult for the user to construct. It is also time-consuming for the user to evaluate each search result. The recommendation techniques having been developed since the early 1990s help solve the problems that traditional IR systems have. This paper explains the basic process and major elements of document recommender systems, especially the two recommendation techniques of content-based filtering and collaborative filtering. Also discussed are the evaluation issue and the problems that current document recommender systems are facing, which need to be taken into account in future system designs. Traditional Information Retrieval (IR) systems have limitations in improving search performance in today’s information environment. The high recall and poor precision of traditional IR systems are only as good as with the accuracy of search query, which is, however, usually difficult for the user to construct. It is also time-consuming for the user to evaluate each search result. The recommendation techniques having been developed since the early 1990s help solve the problems that traditional IR systems have. This paper explains the basic process and major elements of document recommender systems, especially the two recommendation techniques of content-based filtering and collaborative filtering. Also discussed are the evaluation issue and the problems that current document recommender systems are facing, which need to be taken into account in future system designs. 相似文献
13.
协同过滤推荐研究综述 总被引：6，自引：1，他引：6
14.
Borrowing From e-commerce; Are Recommender Systems Good for Libraries? For past 10 years recommender systems have attempted to match customers to materials in the e-commerce world. Utilizing specifically designed recommender systems to meet the needs of patrons, collaborative filtering, and content-based recommender systems are the two basic types with several hybrids created from these two have been significantly enhancing digital library services. Security issues come into play in a special way with libraries, and plagiarism can also affect recommender quality. Open Source tools may be the answer for libraries and their customers, enabling a better utilization of all that a library has to offer. 相似文献
15.
基于信任的电子商务推荐多样性研究 总被引：3，自引：0，他引：3
现有的推荐系统研究大都千方百计地关注于如何提高推荐算法的准确性,考虑到用户兴趣的覆盖范围,这样做的缺陷是只考虑了推荐列表中单个项目的准确度,而忽略了整个推荐列表多样性对用户满意度的影响.近几年的研究表明将信任机制融入到个性化推荐过程中对提高传统协同过滤技术的准确性和鲁棒性有积极的影响,本文提出了基于社会网络信任的推荐多样性算法,该算法通过选择主题多样性好的信任邻居来平衡推荐结果的准确性和多样性.一系列的实验结果表明,该算法能有效地提高推荐的多样性. 相似文献
16.
论数字图书馆个性化信息推荐系统* 总被引：29，自引：3，他引：26
构建一个综合信息检索、信息过滤、数据挖掘等多种技术的信息推荐系统，是数字图书馆实现个性化信息服务的有效手段。阐述了数字图书馆个性化信息推荐系统的目标、功能与资源定位，对推荐系统的结构框架和主要功能作了解剖，并分析了推荐系统实现的主要技术。 相似文献
17.
Vreixo Formoso Diego Fernández Fidel Cacheda Victor Carneiro 《Information Retrieval》2013,16(6):680-696
Collaborative filtering is a popular recommendation technique. Although researchers have focused on the accuracy of the recommendations, real applications also need efficient algorithms. An index structure can be used to store the rating matrix and compute recommendations very fast. In this paper we study how compression techniques can reduce the size of this index structure and, at the same time, speed up recommendations. We show how coding techniques commonly used in Information Retrieval can be effectively applied to collaborative filtering, reducing the matrix size up to 75 %, and almost doubling the recommendation speed. Additionally, we propose a novel identifier reassignment technique, that achieves high compression rates, reducing by 40 % the size of an already compressed matrix. It is a very simple approach based on assigning the smallest identifiers to the items and users with the highest number of ratings, and it can be efficiently computed using a two pass indexing. The usage of the proposed compression techniques can significantly reduce the storage and time costs of recommender systems, which are two important factors in many real applications. 相似文献
18.
Privacy-preserving collaborative filtering algorithms are successful approaches. However, they are susceptible to shilling attacks. Recent research has increasingly focused on collaborative filtering to protect against both privacy and shilling attacks. Malicious users may add fake profiles to manipulate the output of privacy-preserving collaborative filtering systems, which reduces the accuracy of these systems. Thus, it is imperative to detect fake profiles for overall success. Many methods have been developed for detecting attack profiles to keep them outside of the system. However, these techniques have all been established for non-private collaborative filtering schemes. The detection of shilling attacks in privacy-preserving recommendation systems has not been deeply examined. In this study, we examine the detection of shilling attacks in privacy-preserving collaborative filtering systems. We utilize four attack-detection methods to filter out fake profiles produced by six well-known shilling attacks on perturbed data. We evaluate these detection methods with respect to their ability to identify bogus profiles. Real data-based experiments are performed. Empirical outcomes demonstrate that some of the detection methods are very successful at filtering out fake profiles in privacy-preserving collaborating filtering schemes. 相似文献
19.
《Library & information science research》2022,44(4):101191
The use of data mining modern technology in library management systems and information centers is of great importance. With the increasing availability of a large quantity of information, traditional tools and practices without wasting time and cost cannot respond to users accurately and quickly. The present study aims to analyze book circulation transactions and discover the user's book loan patterns to develop a recommender system. The data included 109,639 transactions and information from 8636 user records. Microsoft SQL Server and Matlab software were applied to analyze the data. Item-based collaborative filtering algorithms and decision tree methods were also applied. The results led to the extraction of rules for suggesting books to users. Analysis of the circulation data could be applied to address many issues like evaluation, collection acquisition policies, allocating funding for materials, and suggesting approaches to deselecting and allocating physical space for materials. 相似文献
20.
An Empirical Analysis of Design Choices in Neighborhood-Based Collaborative Filtering Algorithms 总被引：4，自引：0，他引：4
Collaborative filtering systems predict a user's interest in new items based on the recommendations of other people with similar interests. Instead of performing content indexing or content analysis, collaborative filtering systems rely entirely on interest ratings from members of a participating community. Since predictions are based on human ratings, collaborative filtering systems have the potential to provide filtering based on complex attributes, such as quality, taste, or aesthetics. Many implementations of collaborative filtering apply some variation of the neighborhood-based prediction algorithm. Many variations of similarity metrics, weighting approaches, combination measures, and rating normalization have appeared in each implementation. For these parameters and others, there is no consensus as to which choice of technique is most appropriate for what situations, nor how significant an effect on accuracy each parameter has. Consequently, every person implementing a collaborative filtering system must make hard design choices with little guidance. This article provides a set of recommendations to guide design of neighborhood-based prediction systems, based on the results of an empirical study. We apply an analysis framework that divides the neighborhood-based prediction approach into three components and then examines variants of the key parameters in each component. The three components identified are similarity computation, neighbor selection, and rating combination. 相似文献