结合动态多类信息的兴趣点推荐

利用用户的历史签到数据的同时考虑用户的长期偏好和短期偏好已成为当今兴趣点（Point-of-Interest,POI）推荐的主流方法之一。然而,现有方法往往忽略了用户评论中隐含的用户偏好信息,忽视了不同用户在对长期偏好和短期偏好的依赖上所存在的差异性。针对上述局限,本文提出一种结合动态多类信息的兴趣点推荐方法DMGCR。首先,利用注意力机制捕获用户对不同兴趣点的关注程度,定量刻画用户对兴趣点的长期偏好。其次,将评论信息与已有的位置和类别信息相结合,并利用双向长短期记忆网络学习评论文本中隐含的语义特征,在捕获用户对兴趣点情感倾向的基础上准确刻画用户的短期偏好。最后,设计融合动态多类信息的用户偏好综合预测函数,实现下一个兴趣点推荐概率的定量计算。多个数据集上的实验结果验证了该方法在推荐性能上的有效性和优越性。     

Modeling Temporal Dynamics of Users’ Purchase Behaviors for Next Basket Prediction

Next basket prediction attempts to provide sequential recommendations to users based on a sequence of the user’s previous purchases. Ideally, a good prediction model should be able to explore the personalized preference of the users, as well as the sequential relations of the items. This goal of modeling becomes even more challenging when both factors are time-dependent. However, existing methods either take these two aspects as static, or only consider temporal dynamics for one of the two aspects. In this work, we propose the dynamic representation learning approach for time-dependent next basket recommendation, which jointly models the dynamic nature of user preferences and item relations. To do so, we explicitly model the transaction timestamps, as well as the dynamic representations of both users and items, so as to capture the personalized user preference on each individual item dynamically. Experiments on three real-world retail datasets show that our method significantly outperforms several state-of-the-art methods for next basket recommendation.

     

基于图嵌入和GRU的兴趣点推荐模型

下一个兴趣点推荐是基于位置的社交网络(Location-Based Social Network,LBSN)的重要服务之一,其不仅可以帮助用户寻找其感兴趣的目的 地,还能帮助商家提高潜在的收入.目前已有算法提出采用用户行为序列信息以及兴趣点信息进行推荐,但其没有很好地利用兴趣点辅助信息,因此无法缓解冷启动与数据稀疏问题.本文提出了一种基于图嵌入与GRU (Gated Recurrent Unit)的兴趣点推荐模型GE-GRU (Graph Embedding-Gated Recurrent Unit).GE-GR首先通过图嵌入的方法,将兴趣点本身与其辅助信息相融合,得到信息丰富的深层次兴趣点向量,再将其输入到神经网络中,利用GRU对用户近期兴趣偏好进行建模得到用户Embedding表示,最后根据兴趣点排序列表进行下一个兴趣点推荐.本文在一个真实的数据集Foursquare中超过48万条签到记录上进行了实验,采用Accuracy@k指标进行评估,实验结果表明,GE-GRU相比于GRU、LSTM (Long Short-Term Memory)在Accuracy@10上分别有3％和7％的提升.     

融合社交关系和局部地理因素的兴趣点推荐

兴趣点（Point-Of-Interest,POI）推荐是基于位置社交网络（Location-Based Social Network,LBSN）中一项重要的个性化服务,可以帮助用户发现其感兴趣的[POI],提高信息服务质量。针对[POI]推荐中存在的数据稀疏性问题,提出一种融合社交关系和局部地理因素的[POI]推荐算法。根据社交关系中用户间的共同签到和距离关系度量用户相似性,并基于用户的协同过滤方法构建社交影响模型。为每个用户划分一个局部活动区域,通过对区域内[POIs]间的签到相关性分析,建立局部地理因素影响模型。基于加权矩阵分解挖掘用户自身偏好,并融合社交关系和局部地理因素进行[POI]推荐。实验表明,所提出的[POI]推荐算法相比其他方法具有更高的准确率和召回率,能够有效缓解数据稀疏性问题,提高推荐质量。     

基于时空循环神经网络的下一个兴趣点推荐方法

下一个兴趣点推荐已经成为基于位置的社交网络(location-based social networks,LBSNs)中一个重要任务.现有的模型没有深入考虑相邻签到兴趣点之间的转移时空信息,无法对用户访问下一个兴趣点的长短时间偏好和远近距离偏好进行有效建模.本文通过对循环神经网络(recurrent neural ne...     

基于动态和静态偏好的兴趣点推荐算法

针对大多数现有主流兴趣点(POI)推荐算法忽略了融合用户复杂动态偏好和一般静态偏好建模的复杂性问题,提出一个融合复杂动态用户偏好和一般静态用户偏好的POI推荐算法CLSR.首先,在复杂动态偏好建模过程中,基于用户的签到行为及其中的跳过行为设计一个混合神经网络,实现用户的复杂动态兴趣的建模;其次,在一般静态偏好建模过程中...     

Discovery and representation of the preferences of automatically detected groups: Exploiting the link between group modeling and clustering

There are types of information systems, like those that produce group recommendations or a market segmentation, in which it is necessary to aggregate big amounts of data about a group of users in order to filter the data. Group modeling is the process that combines multiple user models into a single model that represents the knowledge available about the preferences of the users in a group. In group recommendation, group modeling allows a system to derive a group preference for each item. Different strategies lead to completely different group models, so the strategy used to model a group has to be evaluated in the domain in which the group recommender system operates. This paper evaluates group modeling strategies in a group recommendation scenario in which groups are detected by clustering the users. Once users are clustered and groups are formed, different strategies are tested, in order to find the one that allows a group recommender system to get the best accuracy. Experimental results show that the strategy used to build the group models strongly affects the performance of a group recommender system. An interesting property derived by our study is that clustering and group modeling are strongly connected. Indeed, the modeling strategy takes the same role that the centroid has when users are clustered, by producing group preferences that are equally distant from the preferences of every user. This “continuity” among the two tasks is essential in order to build accurate group recommendations.     

基于用户隐性反馈行为的下一个购物篮推荐

下一个购物篮推荐是当前电子商务领域中极其重要的一项任务,传统的下一个购物篮推荐方法主要分为时序推荐模型和总体推荐模型。这些方法对点击、收藏、加入购物车等用户的隐性反馈行为利用得不够,并且没有考虑用户行为偏好的时间敏感性。该文提出了一种基于用户隐性反馈行为的下一个购物篮推荐方法,将用户行为按照一定的时间窗口进行划分,对于每个窗口从多个维度抽取用户对商品的时序偏好特征,运用深度学习领域的卷积神经网络模型进行分类器训练。在真实数据集中的实验结果表明,与传统的线性模型和树模型等分类器相比,该文提出的卷积神经网络框架具有较强的特征萃取能力和泛化能力,提高了推荐系统的用户满意度。     

基于社交和地理信息的兴趣点推荐

目前基于用户的协同过滤兴趣点推荐模型认为两个用户之间对彼此的影响是相同的；同时，在计算社交用户相似度时仅仅考虑了用户的朋友集合，未考虑用户住所的地理信息。针对上述问题，提出了一种融合用户、社会和地理信息的兴趣点推荐（Fuse Users、Social and Geographic，FUSG）模型。将非对称用户影响和PageRank算法融入到基于用户的协同过滤算法中，挖掘用户偏好对兴趣点推荐系统的影响；结合社交用户之间的居住距离和用户的共同好友计算用户之间的相似度；利用地理信息挖掘用户签到的地理特征；将改进的协同过滤算法、社交信息与地理信息融合成FUSG模型，进行兴趣点推荐。在真实的数据集上的实验结果表明，FUSG模型不仅能够缓解冷启动问题，且与其他模型相比具有更高的推荐结果。     

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。     

A time-aware spatio-textual recommender system

Location-Based Social Networks (LBSNs) allow users to post ratings and reviews and to notify friends of these posts. Several models have been proposed for Point-of-Interest (POI) recommendation that use explicit (i.e. ratings, comments) or implicit (i.e. statistical scores, views, and user influence) information. However the models so far fail to capture sufficiently user preferences as they change spatially and temporally. We argue that time is a crucial factor because user check-in behavior might be periodic and time dependent, e.g. check-in near work in the mornings and check-in close to home in the evenings. In this paper, we present two novel unified models that provide review and POI recommendations and consider simultaneously the spatial, textual and temporal factors. In particular, the first model provides review recommendations by incorporating into the same unified framework the spatial influence of the users’ reviews and the textual influence of the reviews. The second model provides POI recommendations by combining the spatial influence of the users’ check-in history and the social influence of the users’ reviews into another unified framework. Furthermore, for both models we consider the temporal dimension and measure the impact of time on various time intervals. We evaluate the performance of our models against 10 other methods in terms of precision and recall. The results indicate that our models outperform the other methods.     

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.     

面向推荐系统的分期序列自注意力网络

在推荐系统中,为了充分表达用户反馈数据内部的相互依赖和序列性,准确提取用户的长期/一般偏好、应对数据的动态性,本文提出了一种分期序列自注意力网络(long-term & short-term sequential self-attention network,LSSSAN)进行序列推荐。模型采用自注意力机制和GRU捕捉了用户反馈数据之间的相互依赖和序列性;模型采用注意力机制为不同反馈数据赋予不同权重以动态捕捉重点信息,同时考虑了上下文的动态性;模型基于用户的长期反馈数据,准确表达了用户的长期/一般偏好。该模型在两个数据集上进行训练和测试,结果表明该模型的推荐效果整体优于之前的相关工作。     

A tensor decomposition based collaborative filtering algorithm for time-aware POI recommendation in LBSN

Point of interest (POI) recommendation problem in location based social network (LBSN) is of great importance and the challenge lies in the data sparsity, implicit user feedback and personalized preference. To improve the precision of recommendation, a tensor decomposition based collaborative filtering (TDCF) algorithm is proposed for POI recommendation. Tensor decomposition algorithm is utilized to fill the missing values in tensor (user-category-time). Specifically, locations are replaced by location categories to reduce dimension in the first phase, which effectively solves the problem of data sparsity. In the second phase, we get the preference rating of users to POIs based on time and user similarity computation and hypertext induced topic search (HITS) algorithm with spatial constraints, respectively. Finally the user’s preference score of locations are determined by two items with different weights, and the Top-N locations are the recommendation results for a user to visit at a given time. Experimental results on two LBSN datasets demonstrate that the proposed model gets much higher precision and recall value than the other three recommendation methods.

     

融合知识图谱和轻量图卷积网络的推荐算法

针对知识图谱推荐算法用户端和项目端建模程度不均且模型复杂度较高等问题, 提出融合知识图谱和轻量图卷积网络的推荐算法. 在用户端, 利用用户相似性生成邻居集合, 将用户及其相似用户的交互记录在知识图谱上多次迭代传播, 增强用户特征表示. 在项目端, 将知识图谱中实体嵌入传播, 挖掘与用户喜好相关的项目信息; 接着, 利用轻量图卷积网络聚合邻域特征获得用户和项目的特征表示, 同时采用注意力机制将邻域权重融入实体, 增强节点的嵌入表示; 最后, 预测用户和项目之间的评分. 实验表明, 在Book-Crossing数据集上, 相较于最优基线, AUC和ACC分别提高了1.8%和2.3%. 在Yelp2018数据集上, AUC和ACC分别提高了1.2%和1.4%. 结果证明, 该模型与其他基准模型相比有较好的推荐性能.     

Non-dominated differential context modeling for context-aware recommendations

Context plays an important role in the process of decision making. A user’s preferences on the items may vary from contexts to contexts, e.g., a user may prefer to watch a different type of the movies, if he or she is going to enjoy the movie with partner rather than with children. Context-aware recommender systems, therefore, were developed to adapt the recommendations to different contextual situations, such as time, location, companion, etc. Differential context modeling is a series of recommendation models which incorporate contextual hybrid filtering into the neighborhood based collaborative filtering approaches. In this paper, we propose to enhance differential context modeling by utilizing a non-dominated user neighborhood. The notion of dominance relation was originally proposed in multi-objective optimization, and it was reused to definite non-dominated user neighborhood in collaborative filtering recently. These non-dominated user neighbors refer to the neighbors that dominate others from different perspectives of the user similarities, such as the user-user similarities based on ratings, demographic information, social relationships, and so forth. In this paper, we propose to identify the non-dominated user neighborhood by exploiting user-user similarities over multiple contextual preferences. Our experimental results can demonstrate the effectiveness of the proposed approaches in comparison with popular context-aware collaborative filtering models over five real-world contextual rating data sets.

     

一种基于评分矩阵局部低秩假设融合地理和文本信息的协同排名POI推荐模型*

针对目前LBSN中,用户只对少数兴趣点进行签到,使得用户签到历史数据及其上下文信息（如评论文本）极其稀疏,同时传统的评分推荐系统只考虑用户和评分二元信息,具有一定的局限性。为此,提出一种基于评分矩阵局部低秩假设的局部协同排名兴趣点推荐算法。首先,假设用户-兴趣点矩阵在由用户-兴趣点对所定义度量空间中某些邻域内是低秩;其次,对于地理信息建模采用一种自适应二维核密度方法,然后,对于文本信息利用潜在狄利克雷分配模型挖掘兴趣点相关的文本信息建模用户的兴趣主题;最后,基于局部协同排名模型将兴趣点的地理信息和评论文本信息有效融合。实验结果表明:该模型的性能优于主流先进兴趣点推荐算法。     

融合全局和局部特征的下一个兴趣点推荐方法

随着海量移动数据的积累,下一个兴趣点推荐已成为基于位置的社交网络中的一项重要任务.目前,主流方法倾向于从用户近期的签到序列中捕捉局部动态偏好,但忽略了历史移动数据蕴含的全局静态信息,从而阻碍了对用户偏好的进一步挖掘,影响了推荐的准确性.为此,提出一种基于全局和局部特征融合的下一个兴趣点推荐方法.该方法利用签到序列中的顺序依赖和全局静态信息中用户与兴趣点之间、连续签到之间隐藏的关联关系建模用户移动行为.首先,引入两类全局静态信息,即User-POI关联路径和POI-POI关联路径,学习用户的全局静态偏好和连续签到之间的全局依赖关系.具体地,利用交互数据以及地理信息构建异构信息网络,设计关联关系表示学习方法,利用相关度引导的路径采样策略以及层级注意力机制获取全局静态特征.然后,基于两类全局静态特征更新签到序列中的兴趣点表示,并采用位置与时间间隔感知的自注意力机制来捕捉用户签到序列中签到之间的局部顺序依赖,进而评估用户访问兴趣点概率,实现下一个兴趣点推荐.最后,在两个真实数据集上进行了实验比较与分析,验证了所提方法能够有效提升下一个兴趣点推荐的准确性.此外,案例分析表明,建模显式路径有助于提...     

融合用户偏好和多交互网络的协同过滤算法

推荐系统中的辅助信息可以为推荐提供有用的帮助,而传统的协同过滤算法在计算用户相似度时对辅助信息的利用率低,数据稀疏性大,导致推荐的精度偏低.针对这一问题,本文提出了一种融合用户偏好和多交互网络的协同过滤算法(NIAP-CF).该算法首先根据评分矩阵和项目属性特征矩阵挖掘出用户的项目属性偏好信息,然后使用SBM方法计算用户间的项目属性偏好相似度,并用其改进用户相似度计算公式.在进行评分预测时,构建融合用户-项目属性偏好信息的多交互神经网络预测模型,使用动态权衡参数综合由用户相似度计算出的预测评分和模型的预测评分来进行项目推荐.本文使用MovieLens数据集进行实验验证,实验结果表明改进算法能够提高推荐的精度,降低评分预测的MAE和RMSE值.     

Introduction: In-Use,In-Situ: Extending Field Research Methods—Part 2

This study used protocol analyses and user drawings of their models of the system to investigate the “getting lost” problem in hypertext navigation. The “getting lost” problem is viewed as occurring when routine expectations of naive users, concerning appropriate linear sequences, are violated. Several ways in which users persistently attempt to work within a linear model, despite its inapplicability, are examined. The transition to more hierarchical user models is described.