CAESAR: context-aware explanation based on supervised attention for service recommendations

Explainable recommendations have drawn more attention from both academia and industry recently, because they can help users better understand recommendations (i.e., why some particular items are recommended), therefore improving the persuasiveness of the recommender system and users’ satisfaction. However, little work has been done to provide explanations from the angle of a user’s contextual situations (e.g., companion, season, and destination if the recommendation is a hotel). To fill this research gap, we propose a new context-aware recommendation algorithm based on supervised attention mechanism (CAESAR), which particularly matches latent features to explicit contextual features as mined from user-generated reviews for producing context-aware explanations. Experimental results on two large datasets in hotel and restaurant service domains demonstrate that our model improves recommendation performance against the state-of-the-art methods and furthermore is able to return feature-level explanations that can adapt to the target user’s current contexts.

     

垂直学习社区基于学习兴趣与风格的社会化推荐算法

垂直学习社区包含了海量的学习资源,出现了信息过载现象,个性化推荐是解决这个难题的方法之一.但垂直学习社区中评分数据稀疏而文本、社交信息丰富,传统的协同过滤推荐算法不完全适用.基于用户产生的文本和行为信息,利用作者主题模型构建新的用户学习兴趣相似度衡量模型;根据用户交互行为信息综合考虑信任与不信任因素构建用户全面信任关系计算全面信任度;通过分析用户多维度学习行为模式,自动识别用户学习风格;最后提出融合兴趣相似度、全面信任度及学习风格的社会化推荐算法.用垂直学习社区网站CSDN实际数据集进行了实验分析.结果表明本文提出的推荐方法能更好向用户推荐其感兴趣的学习资源,有效地提高了推荐精度,进而提高用户学习效果.     

A Robust Missing Data-Recovering Technique for Mobility Data Mining

Based on location information, users’ mobility profile building is the main task for making different useful systems such as early warning system, next destination and route prediction, tourist guide, mobile users’ behavior-aware applications, and potential friend recommendation. For mobility profile building, frequent trajectory patterns are required. The trajectory building is based on significant location extraction and the user’s actual movement prediction. Previous works have focused on significant places extraction without considering the change in GSM (global system for mobile communication) network and is based on complete data analysis. Since network operators change the GSM network periodically, there are possibilities of missing values and outliers. These missing values and outliers must be addressed to ensure actual mobility and for the efficient extraction of significant places, which are the basis for users’ trajectory building. In this paper, we propose a methodology to convert geo-coordinates into semantic tags and we also purposed a clustering methodology for recovering missing values and outlier detection. Experimental results prove the efficiency and effectiveness of the proposed scheme.     

深度学习方法在兴趣点推荐中的应用研究综述

在基于位置的社交网络（LBSN）中,用户可以在兴趣点（POI）进行签到以记录行程,也可以与其他用户分享自身的感受并形成社交好友关系。POI推荐是LBSN提供的一项重要服务,其可以帮助用户快速发现感兴趣的POI,也有利于POI提供商更全面地了解用户偏好,并有针对性地提高服务质量。POI推荐主要基于对用户历史签到数据以及用户生成内容、社交关系等信息的分析来实现。系统归纳POI推荐中所面临的时空序列特征提取、内容社交特征提取、多特征整合、数据稀疏性问题处理这4个方面的挑战,分析在POI推荐中使用深度学习方法解决上述问题时存在的优势以及不足。在此基础上,展望未来通过深度学习提高POI推荐效果的研究方向,即通过增量学习加速推荐模型更新、使用迁移学习缓解冷启动问题以及利用强化学习建模用户动态偏好,从而为实现效率更高、用户体验质量更好的推荐系统提供新的思路。     

基于自监督学习的社交网络用户轨迹预测模型

针对当前用户轨迹数据建模中存在的签到点稀疏性、长时间依赖性和移动模式复杂等问题,提出基于自监督学习的社交网络用户轨迹预测模型SeNext,对用户轨迹进行建模和训练来预测用户的下一个兴趣点（POI）。首先,使用数据增强的方式来丰富训练数据样本,以解决数据不足及个别用户足迹太少导致的模型泛化能力不足的问题;其次,将循环神经网络（RNN）、卷积神经网络（CNN）和注意力机制分别用于当前轨迹和历史轨迹的建模中,以此从高维稀疏的数据中提取有用的表示,用来匹配用户过去最相似的移动方式。最后,通过结合自监督学习并引入对比损失优化噪声对比估计（InfoNCE）,SeNext在潜在空间学习隐含表示来预测用户的下一个POI。实验结果表明,在纽约数据集上,SeNext比最新的VANext（Variational Attention based Next）模型的预测准确度在Top@1上提高了11.10%左右。     

Improving Friend Recommendation for Online Learning with Fine-Grained Evolving Interest

Friend recommendation plays a key role in promoting user experience in online social networks (OSNs). However, existing studies usually neglect users’ fine-grained interest as well as the evolving feature of interest, which may cause unsuitable recommendation. In particular, some OSNs, such as the online learning community, even have little work on friend recommendation. To this end, we strive to improve friend recommendation with fine-grained evolving interest in this paper. We take the online learning community as an application scenario, which is a special type of OSNs for people to learn courses online. Learning partners can help improve learners’ learning effect and improve the attractiveness of platforms. We propose a learning partner recommendation framework based on the evolution of fine-grained learning interest (LPRF-E for short). We extract a sequence of learning interest tags that changes over time. Then, we explore the time feature to predict evolving learning interest. Next, we recommend learning partners by fine-grained interest similarity. We also refine the learning partner recommendation framework with users’ social influence (denoted as LPRF-F for differentiation). Extensive experiments on two real datasets crawled from Chinese University MOOC and Douban Book validate that the proposed LPRF-E and LPRF-F models achieve a high accuracy (i.e., approximate 50% improvements on the precision and the recall) and can recommend learning partners with high quality (e.g., more experienced and helpful).     

A system for destination and future route prediction based on trajectory mining

Location is a key context ingredient and many existing pervasive applications rely on the current locations of their users. However, with the ability to predict the future location and movement behavior of a user, the usability of these applications can be greatly improved. In this paper, we propose an approach to predict both the intended destination and the future route of a person. Rather than predicting the destination and future route separately, we have focused on making prediction in an integrated way by exploiting personal movement data (i.e. trajectories) collected by GPS. Since trajectories contain daily whereabouts information of a person, the proposed approach first detects the significant places where the person may depart from or go to using a clustering-based algorithm called FBM (Forward–Backward Matching), then abstracts the trajectories based on a space partitioning method, and finally extracts movement patterns from the abstracted trajectories using an extended CRPM (Continuous Route Pattern Mining) algorithm. Extracted movement patterns are organized in terms of origin–destination couples. The prediction is made based on a pattern tree built from these movement patterns. With the real personal movement data of 14 participants, we conducted a number of experiments to evaluate the performance of our system. The results show that our approach can achieve approximately 80% and 60% accuracy in destination prediction and 1-step prediction, respectively, and result in an average deviation of approximately 60 m in continuous future route prediction. Finally, based on the proposed approach, we implemented a prototype running on mobile phones, which can extract patterns from a user’s historical movement data and predict the destination and future route.     

图表示学习短视频智能推荐研究综述

随着短视频数量的爆发式增长, 精准的个性化短视频推荐成为学术界和工业界的迫切需求。然而,现有的推荐方法没有考虑实际的短视频具有数据多源异构多模态、用户行为复杂多样、用户兴趣动态变化等特点。短视频模态间的语义鸿沟、社交网络用户多行为挖掘、用户动态兴趣捕捉依然是短视频推荐领域面临的三个重要问题。针对当前推荐系统存在的问题,并充分考虑短视频推荐系统的实际需求,本文介绍了短视频推荐中基于图表示学习的短视频推荐方法;研究了短视频异构多模态特征表示,充分挖掘视频内容特征并进行高效融合;研究了短视频社交网络用户多行为表示,通过社交网络用户多种行为挖掘更细粒度的用户偏好;研究了用户的动态偏好表示方法,通过利用时序信息建模用户的动态兴趣,保证推荐结果的准确度并增加其多样性与个性化。本研究可在理论和实践上推进基于图特征学习的短视频推荐研究,也可作为短视频推荐系统的关键技术。     

Learning influence from heterogeneous social networks

Influence is a complex and subtle force that governs social dynamics and user behaviors. Understanding how users influence each other can benefit various applications, e.g., viral marketing, recommendation, information retrieval and etc. While prior work has mainly focused on qualitative aspect, in this article, we present our research in quantitatively learning influence between users in heterogeneous networks. We propose a generative graphical model which leverages both heterogeneous link information and textual content associated with each user in the network to mine topic-level influence strength. Based on the learned direct influence, we further study the influence propagation and aggregation mechanisms: conservative and non-conservative propagations to derive the indirect influence. We apply the discovered influence to user behavior prediction in four different genres of social networks: Twitter, Digg, Renren, and Citation. Qualitatively, our approach can discover some interesting influence patterns from these heterogeneous networks. Quantitatively, the learned influence strength greatly improves the accuracy of user behavior prediction.     

近似到达时间约束下的语义轨迹频繁模式挖掘

随着GPS定位技术的不断发展与智能移动设备的普及,轨迹数据的获取变得越来越容易,同时,轨迹数据相关应用的需求也逐渐增多.在轨迹数据上加入语义信息,可以得到体积较小、质量较高、能够更好地反映用户行为的语义轨迹,在其上实现旅游线路推荐、路线预测、用户生活模式挖掘、朋友推荐等应用,可以更好地满足用户需求.挖掘语义轨迹的频繁模式是实现这些应用的技术基础,而在很多情况下,用户对语义轨迹频繁模式常存在到达时间方面的需求,比如按特定时间游玩热门景点的同时需要按时到达车站候车.现有的语义轨迹模式挖掘方法大多没有考虑到达时间的约束,挖掘出的频繁模式缺少到达时间信息;少数方法考虑了精确的到达时间,但因为约束太强会导致无法挖掘到频繁的模式.因此,首次对近似到达时间约束下的语义轨迹频繁模式（approximate arrival-time constrained frequent pattern,简称AAFP）挖掘方法进行了研究,并给出了其形式化定义;通过时间轴划分提出了挖掘AAFP的基线算法,并通过建立索引AAP-tree提出了改进后的高效、灵活的AAFP挖掘算法;之后提出了信息熵增量公式,并给出了时间轴划分及AAP-tree的高效维护方法;最后在真实数据集上进行实验,验证了方法的有效性及高效性.     

融合深度神经网络和张量分解的地点推荐算法

为提高利用张量分解技术进行基于位置社交网络的地点推荐的推荐性能,提出一种利用张量分解技术且融合神经网络的地点推荐算法。融合多层感知机和长短期记忆网络基于张量分解技术建模用户的签到行为,将学习到的用户偏好表示馈送到推荐生成器和推荐判别器组成的对抗生成网络中,通过对抗训练学习最佳用户偏好表示用于推荐。基于真实数据集的实验验证了该算法的有效性和高效性。     

基于轨迹相似度的伴随人员推荐

移动网络和智能终端的发展使得基于优质用户的伴随人员的推荐成为互联网发展的热点之一,而伴随人员的推荐算法则是至关重要的因素.针对以往基于地理位置的用户轨迹性相似推荐算法中需基于地理位置或基站数据,且数据稀疏时推荐结果不理想的问题,提出了基于IP场所的轨迹余弦相似度的伴随人员推荐,以更完善的IP场所数据代替地理位置数据,以一段时间的纵向日期和横向时刻分别计算余弦相似度以消除数据稀疏性问题.最后推荐出了相似度质量更高的伴随人员.     

Product recommendation with temporal dynamics

In many E-commerce recommender systems, a special class of recommendation involves recommending items to users in a life cycle. For example, customers who have babies will shop on Diapers.com within a relatively long period, and purchase different products for babies within different growth stages. Traditional recommendation algorithms produce recommendation lists similar to items that the target user has accessed before (content filtering), or compute recommendation by analyzing the items purchased by the users who are similar to the target user (collaborative filtering). Such recommendation paradigms cannot effectively resolve the situation with a life cycle, i.e., the need of customers within different stages might vary significantly. In this paper, we model users’ behavior with life cycles by employing hand-crafted item taxonomies, of which the background knowledge can be tailored for the computation of personalized recommendation. In particular, our method first formalizes a user’s long-term behavior using the item taxonomy, and then identifies the exact stage of the user. By incorporating collaborative filtering into recommendation, we can easily provide a personalized item list to the user through other similar users within the same stage. An empirical evaluation conducted on a purchasing data collection obtained from Diapers.com demonstrates the efficacy of our proposed method.     

一种潜在特征同步学习和偏好引导的推荐方法

根据用户的历史评分数据为用户提供推荐的商品列表，是目前推荐系统研究的主流.研究者发现，随着用户参与度的不断提高，将反映用户偏好的评论文本与评分数据结合，可以进一步提高推荐的质量.提出了基于潜在特征同步学习和偏好引导的商品推荐方法，将评论文本的主题与用户的"打分偏好"进行关联，同步学习用户评论文本的潜在主题、评分矩阵的用户潜在因子和商品潜在因子，并将潜在主题作为用户个人偏好引导来约束推荐方法对商品的预测打分.该方法对推荐质量的优化主要体现在两个方面：一是在评论文本的潜在主题和评分数据的两种潜在因子之间建立映射关系，同步求解主题模型和矩阵分解模型；二是将从评论文本中学习得到的潜在主题作为用户对商品的个性偏好引入到矩阵分解中，进一步优化推荐方法.在来自Amazon网站的28组真实数据集上进行实验，以均方误差为评价指标，与已有的模型进行了对比分析.实验结果表明，该方法有效减少了推荐误差，与已有的TopicMF方法相比，均方误差在数据子集上最大减少了3.32%，平均减少了0.92%.     

Topic-tracking-based dynamic user modeling with TV recommendation applications

One of the challenging issues in TV recommendation applications based on implicit rating data is how to make robust recommendation for the users who irregularly watch TV programs and for the users who have their time-varying preferences on watching TV programs. To achieve the robust recommendation for such users, it is important to capture dynamic behaviors of user preference on watched TV programs over time. In this paper, we propose a topic tracking based dynamic user model (TDUM) that extends the previous multi-scale dynamic topic model (MDTM) by incorporating topic-tracking into dynamic user modeling. In the proposed TDUM, the prior of the current user preference is estimated as a weighted combination of the previously learned preferences of a TV user in multi-time spans where the optimal weight set is found in the sense of the evidence maximization of the Bayesian probability. So, the proposed TDUM supports the dynamics of public users’ preferences on TV programs for collaborative filtering based TV program recommendation and the highly ranked TV programs by similar watching taste user group (topic) can be traced with the same topic labels epoch by epoch. We also propose a rank model for TV program recommendation. In order to verify the effectiveness of the proposed TDUM and rank model, we use a real data set of the TV programs watched by 1,999 TV users for 7 months. The experiment results demonstrate that the proposed TDUM outperforms the Latent Dirichlet Allocation (LDA) model and the MDTM in log-likelihood for the topic modeling performance, and also shows its superiority compared to LDA, MDTM and Bayesian Personalized Rank Matrix Factorization (BPRMF) for TV program recommendation performance in terms of top-N precision-recall.     

融合时间序列的POI动态推荐算法

兴趣点(POI)的签到数据体现了用户的偏好和兴趣点的分布特征,这在兴趣点推荐领域有极为重要的价值.为了缓解数据稀疏造成的推荐不准确等问题,本文提出了融合时间序列的POI动态推荐算法,结合用户与用户之间的关系、兴趣点位置以及流行度信息等.首先划分时间序列,得到时间因子的相似度;其次时间序列融入到基于用户的协同过滤算法,再根据时间的连续性特征得到基于用户的预测评分,然后将地理影响因子与基于时间的流行度信息结合,预测用户的评分,进而与基于用户的评分加权融合;最后,在Gowalla数据集上进行实验,结果表明,本文提出的融合时间序列的POI动态推荐算法能够有效减小推荐误差,提高推荐精度与召回率.     

深度融合辅助信息的跨域推荐算法

协同过滤已成功用于为用户提供个性化的产品和服务,然而它面临数据稀疏和冷启动的问题。一种解决方案是结合辅助信息,另一种是从相关领域学习知识。综合考虑了这两个方面,提出一种深度融合辅助信息的跨域推荐算法CICDR,它集成了集体矩阵分解和深度迁移学习。该算法通过Semi-SDAE和矩阵分解（MF）在源域和目标域中进行建模,学习评分信息和辅助信息中的有效特征向量,并利用用户的隐式反馈信息来做出更准确的推荐。通过这种方式,在两个领域中学习到的用户和项目潜在因素为推荐保留了更多的语义信息。通过非完备正交非负矩阵三分解（IONMTF）产生桥接两个相关领域的公共潜在因素,以缓解目标域中的冷启动和数据稀疏问题。在三个真实数据集上与四种经典算法进行对比,验证了提出算法的有效性,进一步提高了推荐精度和用户满意度。     

基于区域划分的轨迹隐私保护方法

针对现有k匿名方法易受连续查询攻击以及在用户数稀少时难以构建匿名区域问题,提出一种基于区域划分的轨迹隐私保护方法。查询用户利用第三方辅助服务器获得拥有特定区域历史查询点的用户组,并通过P2P协议获得用户组中用户的历史查询点,从中搜索所需的查询结果,以提高查询效率。另外,该方法通过发送伪查询点迷惑攻击者,以及利用覆盖用户真实轨迹的区域划分方法,将多个查询点隐藏在同一子区域中,使攻击者无法重构用户的真实轨迹,以保证安全性。实验结果表明,所提方法随着偏离距离和缓存时间的增大,用户轨迹隐私的安全性会提高。在用户数为1500时,与协作轨迹隐私保护（CTPP）方法相比,安全性平均提高约50%,查询效率平均提高约35%（子区域数为400）。     

基于注意力机制的协同卷积动态推荐网络

一直以来, 各种推荐系统关注于如何挖掘用户与物品特征间的潜在关联, 特征信息的充分利用有利于用户到物品的精准匹配. 基于矩阵分解和分解机的推荐算法是该领域的主流, 前者学习用户历史行为而后者分析对象特征关系, 但都难以兼顾用户行为与个体特征. 而近年来, 深度神经网络凭借其强大的特征学习能力和灵活可变的结构被应用到了推荐系统领域. 鉴于此, 本文提出了一种基于注意力机制的协同卷积动态推荐网络(Attentionbased collaborative convolutional dynamic network, ACCDN), 它通过注意力机制实现用户历史行为、用户画像与物品属性的多重交互, 再通过卷积网络逐层捕捉更高阶的特征交互. 网络同时接受不同组块输出的低阶至高阶信息, 最后给出用户对指定物品青睐评分概率的预估. 而且本文还提出了一种基于无参时间衰减的用户兴趣标签来量化用户关注的变化. 通过比较若干先进模型在两个现实数据集的表现, 本文设计的动态推荐模型不但能够缓解推荐时滞性, 还能明显提高推荐质量, 为用户带来更好的个性化服务体验.