Mining user preferences of new locations on location-based social networks: a multidimensional cloud model approach

In recent years, the prevalent of location-based social networks contributes massive data for location recommendation. Although collaborative filtering (CF) algorithm has been widely employed for location recommendation, it suffers the data sparsity and the high time complexity as it estimates the similarity of users by the common locations. In this paper, we extend the two-dimensional cloud model to the multidimensional cloud model and utilize it to the measure the similarity of user preferences and user behaviors. This method not only considers the multiple attributes of users (e.g., the diversity of user preferences), but also alleviates the sparsity of location recommendation based on CF algorithm to some extent. Then we integrate the similarity of user preferences, social ties and user behaviors into CF algorithm, which is expected to mine user preferences of new locations (MUPNL) more precisely. Furthermore, in order to improve the efficiency of the MUPNL algorithm, we parallelize it with Mapreduce framework. Experimental results on Yelp academic dataset demonstrate the good performance of the distributed MUPNL algorithm in accuracy and efficiency.     

移动网络服务中基于认知心理学的用户偏好提取方法

 迅速增长的移动网络服务给人们带来沉重的移动信息负担.移动用户偏好提取方法是缓解"移动信息过载"问题的有效手段.受加工水平模型和分布式认知理论的启发,提出一种基于认知心理学的移动用户偏好提取方法.在移动用户偏好信息结构建模的基础上,引入服务加工水平认知、有效上下文认知的概念,并计算其对用户偏好提取的影响,然后分别提取基于服务加工水平认知和基于有效上下文认知的用户偏好,最终提取综合的用户偏好.实验结果表明,该方法能有效提高移动用户偏好提取精确度,为用户提供满足个性化需求的移动网络服务.     

一种基于动态角色标识和张量分解的推荐模型

社交网络推荐中,通常未依据用户兴趣变化进行用户角色动态标注,会造成推荐预测误差,并且用户评分数据稀疏造成评分预测不准确.根据以上问题,本文提出一种基于动态角色标识和张量分解的推荐模型.首先,针对用户角色无差别标识问题,引入信息熵指标度量用户兴趣多样性,对目标用户进行角色定量标识.其次,考虑到用户兴趣漂移现象,提出基于时间窗的动态角色标识方式,解决静态角色标识产生的个体评分数据无偏好差异问题,实现用户评分数据层次化处理.最后,为提高评分预测准确率,通过引入张量分解在数据维度转换和数据压缩的特性,构建基于"用户-项目-角色"张量分解的评分预测模型.同时,在张量分解的过程中,通过对数据缺失值处理,提高评分预测精度,生成目标用户推荐列表.实验表明,该方法缓解了用户无角色差异形成的预测误差问题,并能够有效改善评分数据稀疏情况下传统方法推荐精度不足的问题,提高推荐效率.     

基于移动用户上下文相似度的协同过滤推荐算法

该文面向移动通信网络领域的个性化服务推荐问题,通过将移动用户上下文信息引入协同过滤推荐过程,提出一种基于移动用户上下文相似度的改进协同过滤推荐算法。该算法首先计算基于移动用户的上下文相似度,以构造目标用户当前上下文的相似上下文集合,然后采用上下文预过滤推荐方法对移动用户-移动服务-上下文3维模型进行降维得到移动用户-移动服务2维模型,最后结合传统2维协同过滤算法进行偏好预测和推荐。仿真数据集和公开数据集实验表明,该算法能够用于移动网络服务环境下的用户偏好预测,并且与传统协同过滤相比具有更高的推荐精确度。     

基于类别转移加权张量分解模型的兴趣点分区推荐

基于位置社交网络的兴趣点(POI)推荐是人们发现有趣位置的重要途径,然而,现实中用户在不同区域的地点偏好侧重的差异,加之高维度的历史签到信息,使得精准而又个性化的POI推荐极富挑战性.对此,该文提出一种新型的基于类别转移加权张量分解模型的兴趣点分区推荐算法(WTD-PR).通过结合用户连续行为和时间特征,来充分利用用户...     

融合多权重因素的低秩概率矩阵分解推荐模型

针对个性化推荐精度较低、对冷启动敏感等问题,该文提出一种融合多权重因素的低秩概率矩阵分解推荐模型MWFPMF.模型利用给定的社交网络构建信任网络,借助Page rank算法和信任传递机制求取用户间信任度;基于Page rank计算用户社会地位,利用活动评分和评分时间修正用户间关系权重;引入词频-逆文本频率技术(TF-I...     

基于用户行为认知的在线社交网络协同推荐

近年来,在线社交网络成为人们工作、生活不可或缺的信息共享与交流工具,如何对海量庞杂、大范围时空关联的用户行为信息进行认知并据此提供个性化的推荐服务,已成为在线社交网络发展重点关注的问题。为此,提出了一种基于用户行为认知的在线社交网络协同推荐框架,在对用户特征、文本信息及兴趣偏好等行为进行认知的基础上,利用协同过滤算法,实现个性化的推荐服务。实验结果验证了提出的基于用户行为认知的协同推荐策略具有较好的稳定性和实际应用效果。     

基于模糊权重相似性的协同过滤算法研究

在传统协同过滤算法中,相似度直接依据用户评分。但是,用户评分会受各种不确定因素影响。采用数值评分的推荐系统收集到的用户喜好信息是模糊、不精确和不完整的。单一的数值不能包含丰富的信息来表达用户喜好,也会导致推荐结果的不准确性。文中定义了几种模糊集的隶属函数,提出了基于模糊逻辑的相似度计算方法。实验结果表明,基于模糊权重的相似度有效的提高了推荐系统的预测准确度,一定程度上解决了协同过滤算法的可扩展性和数据稀疏性问题。     

基于置信用户偏好模型的电视推荐系统

基于用户偏好的电视节目个性化推荐是一种内容的推荐算法。其中用户偏好的不确定性和描述上的模糊性是用户模型建立的难点。在此首先通过对样本用户过往观看记录数据进行分析,发现用户偏好存在一定的时不变性。把偏好在一定时间内不发生变化的用户称作置信用户,在这个基础上,建立基于节目特征向量空间的用户偏好模型,并提出基于用户偏好度模型的推荐算法。该算法通过用户观看视频的历史记录得到用户的偏好模型,并基于该偏好模型向用户推荐节目。仿真实验证明了算法的收敛性和有效性。     

Novelty‐driven recommendation by using integrated matrix factorization and temporal‐aware clustering optimization

With the rapid proliferation of information and communication technology (ICT), the vast amount of available data creates information overload. The Websites and e‐commerce applications employ several information filtering methods such as personalized recommender system to manage the information overload. The recommender system assists the users in obtaining the desired list of products based on their interest. Several existing research works focus on the novelty or unexpectedness in the recommendation list while ensuring the quality to enhance the recommendation mechanism. It is essential to balance the unexpected and useful products or services to generate the satisfactory personalized recommendations with novelty. Thus, this paper proposes a novelty‐driven movie suggestion using integrated matrix factorization and temporal‐aware clustering optimization (NOMINATE). The proposed approach determines the personalized preferences through probabilistic matrix factorization (PMF) and contextually updates the rules and extracts the user preferences based on the inherent features of both the users and movies with temporal information. The NOMINATE approach also suggests the novelty‐driven, and desired top‐N movies to the users through the K‐means, and particle swarm optimization (PSO)‐based clustering algorithm with the help of LOD source. To identify the expert users, the NOMINATE approach applies the K‐means and PSO‐based clustering algorithm to enrich the personalized features of the users. Moreover, it integrates the relevant features with the preferred set of features for each user using the LOD source and decides a set of optimal preferences of the users. Finally, the NOMINATE approach generates the top‐N recommendation list for the corresponding user through ranking method. The experiment results stipulate that the NOMINATE approach personalize the top‐N movie recommendations with high performance regarding accuracy and novelty when compared with the existing recommendation method.     

基于簇级匈牙利与联盟博弈联合的CF mMIMO导频分配算法

针对无小区大规模MIMO（Cell-Free Massive Multiple-Input Multiple-Output,CF mMIMO）系统,提出了一种基于簇级匈牙利与联盟博弈联合的导频分配算法。首先,利用用户的地理位置信息将系统中的用户划分为多个用户簇并为相同簇内的用户分配相互正交的导频序列,从空间角度减轻用户复用导频造成的导频污染;然后,在此基础上利用改进收益矩阵的匈牙利算法优化导频分配方案;最后,利用联盟博弈的思想避免因为固定的导频使用次数而带来的性能上限。仿真结果表明,该导频分配算法性能优于多种典型导频分配算法。     

基于上下文相似度和社会网络的移动服务推荐方法

针对传统的基于协同过滤的移动服务推荐方法存在的数据稀疏性和用户冷启动问题,提出一种基于上下文相似度和社会网络的移动服务推荐方法(Context-similarity and Social-network based Mobile Service Recommendation,CSMSR).该方法将基于用户的上下文相似度引入个性化服务推荐过程,并挖掘由移动用户虚拟交互构成的社会关系网络,按照信任度选取信任用户;然后结合基于用户评分相似度计算发现的近邻,分别从相似用户和信任用户中选择相应的邻居用户,对目标用户进行偏好预测和推荐.实验表明,与已有的服务推荐方法TNCF、SRMTC及CF-DNC相比,CSMSR方法有效地缓解数据稀疏性并提高推荐准确率,有利于发现用户感兴趣的服务,提升用户个性化服务体验.     

Location privacy preserving scheme against attack from friends in SNS

In order to enrich the performance of the user's location information and to meet the diverse needs of users,a location privacy protection scheme based on attribute encryption was designed,which provided precise,more accurate,fuzzy and private four modes to manage the location information.The scheme was based on the algorithm of WT-CP-ABE.The location information was divided into three parts according to a close friend of grade,then the key infor-mation and position information was encrypted with attribute-based encryption and symmetric encryption method respec-tively and the ciphertext was published to the social network.The security of the scheme is analyzed,which shows that the scheme has the advantage of user attribute information confidentiality,data confidentiality and can resist the collusion attack.     

Context‐Aware Active Services in Ubiquitous Computing Environments

With the advent of ubiquitous computing environments, it has become increasingly important for applications to take full advantage of contextual information, such as the user's location, to offer greater services to the user without any explicit requests. In this paper, we propose context‐aware active services based on context‐aware middleware for URC systems (CAMUS). The CAMUS is a middleware that provides context‐aware applications with a development and execution methodology. Accordingly, the applications based on CAMUS respond in a timely fashion to contextual information. This paper presents the system architecture of CAMUS and illustrates the content recommendation and control service agents with the properties, operations, and tasks for context‐aware active services. To evaluate CAMUS, we apply the proposed active services to a TV application domain. We implement and experiment with a TV content recommendation service agent, a control service agent, and TV tasks based on CAMUS. The implemented content recommendation service agent divides the user's preferences into common and specific models to apply other recommendations and applications easily, including the TV content recommendations.     

The Adaptive Ontology-Based Personalized Recommender System

Recommender systems provide strategies that help users search or make decisions within the overwhelming information spaces nowadays. They have played an important role in various areas such as e-commerce and e-learning. In this paper, we propose a hybrid recommendation strategy of content-based and knowledge-based methods that are flexible for any field to apply. By analyzing the past rating records of every user, the system learns the user’s preferences. After acquiring users’ preferences, the semantic search-and-discovery procedure takes place starting from a highly rated item. For every found item, the system evaluates the Interest Intensity indicating to what degree the user might like it. Recommender systems train a personalized estimating module using a genetic algorithm for each user, and the personalized estimating model helps improve the precision of the estimated scores. With the recommendation strategies and personalization strategies, users may have better recommendations that are closer to their preferences. In the latter part of this paper, a real-world case, a movie-recommender system adopting proposed recommendation strategies, is implemented.     

基于深度学习的混合兴趣点推荐算法

针对现有兴趣点推荐的初始化和忽视评论信息语义上下文信息的问题,将深度学习融入推荐系统中已经成为兴趣点推荐研究的热点之一。该文提出一种基于深度学习的混合兴趣点推荐模型(MFM-HNN)。该模型基于神经网络融合评论信息与用户签到信息来提高兴趣点推荐的性能。具体地,利用卷积神经网络学习评论信息的特征表示,利用降噪自动编码对用户签到信息进行初始化。进而,基于扩展的矩阵分解模型融合评论信息特征和用户签到信息的初始值进行兴趣点推荐。在真实签到数据集上进行实验,结果表明所提MFM-HNN模型相比其他先进的兴趣点推荐具有更好的推荐性能。     

基于时空上下文信息的POI推荐模型研究

随着基于位置的社交网络(LBSN)技术的快速发展,为移动用户提供个性化服务的兴趣点(POI)推荐成为关注重点。由于POI推荐面临着数据稀疏、影响因素多和用户偏好复杂的挑战,因此传统的POI推荐往往只考虑签到频率以及签到时间和地点对用户的影响,而忽略了签到序列中用户前后行为的关联影响。为了解决上述问题,该文通过序列的表示考虑签到数据的时间影响和空间影响,建立了时空上下文信息的POI推荐模型(STCPR),为POI推荐提供了更精准的个性化偏好。该模型基于序列到序列的框架下,将用户信息、POI信息、类别信息和时空上下文信息进行向量化后嵌入GRU网络中,同时利用了时间注意力机制、全局和局部的空间注意力机制来综合考虑用户偏好与变化趋势,从而向用户推荐感兴趣的Top-N的POI。该文通过在两个真实的数据集上实验来验证模型的性能。实验的结果表明,该文所提出的方法在召回率(Recall)和归一化折损累计增益(NDCG)方面优于几种现有的方法。     

Dynamic resource allocation in mobile heterogeneous cellular networks

User mobility is a challenging issue in macro and femto cellular networks for the fifth-generation and newer mobile communications due to the time-varying interference and topology experienced. In this paper, we consider an OFDMA-based two-tier network with one macro cell and several femto cells, wherein each macro user and/or femto user can leave or enter its serving cell frequently, referred to as user mobility. A resource allocation problem with different rate requirements of mobile users is then formulated. Assuming well knowledge of the user locations and the channel state information, we propose a dynamic algorithm with static and dynamic parts for a better trade-of between computational complexity and system throughput. The static algorithm, named interference weighted cluster algorithm in this paper, is based on the graph theory to cluster the femtocells by minimizing the interference between clusters, while the dynamic algorithm is to deal with the user mobility by sharing the resource blocks under the constraints of rate requirements. Numerical results are demonstrated to show the effectiveness of the proposed dynamic resource allocation algorithm in terms of capacity, computational time, and outage probability.

     

A User-Centered Approach to Active Replica Management in Mobile Environments

A replication scheme determines the number and location of replicas in a distributed system. Traditional static replication schemes do not perform well in mobile environments due to the assumptions of fixed hosts, relatively static access patterns, and the lack of considerations on users' behalf. For effective data management in mobile environments, we define the notion of activity-data dependency for inferring the information requirement of mobile users based on their scheduled activities. We then propose a dynamic replication scheme which employs user profiles for recording users' mobility schedules, accesses behaviors and read/write patterns, and actively reconfigures the replicas to adapt to changes in user locations, data requests, and system status. Simulation results demonstrate that the scheme can accurately predict the data requirement to facilitate effective replication, reduce response time, and increase data availability     

基于用户相似度的协同过滤推荐算法简

协同过滤推荐算法通过研究用户的喜好,实现从海量数据资源中为用户推荐其感兴趣的内容,在电子商务中得到了广泛的应用。然而,当此类算法应用到社交网络时,传统的评价指标与相似度计算的重点发生了变化,从而出现推荐算法效率偏低,推荐准确度下降问题,导致社交网络中用户交友推荐满意度偏低。针对这一问题,引入用户相似度概念,定义社交网络中属性相似度,相似度构成与计算方法,提出一种改进的协同过滤推荐算法,并给出推荐质量与用户满意度评价方法。实验结果表明：改进算法能有效改善社交网络中的推荐准确性并提高推荐效率,全面提高用户满意度。