论文下载地址: https://arxiv.org/abs/2002.00844v2
发表期刊:TKDE
Publish time: 2020
作者及单位:
数据集: 正文中的介绍
代码:
其他:
其他人写的文章
简要概括创新点: 作者们去年发了DiffNet(只考虑的Social Network),今年再发DiffNet++(再考虑上user-item interset Network).
- (1)we propose DiffNet++, an improved algorithm of DiffNet that models the neural influence diffusion and interest diffusion in a unified framework. (我们提出了DiffNet++,这是一种改进的DiffNet算法,它在一个统一的框架中对神经影响扩散和兴趣扩散进行建模。)
- (2)By reformulating the social recommendation as a heterogeneous graph with social network and interest network as input, DiffNet++ advances DiffNet by injecting both the higher-order user latent interest reflected in the user-item graph and higher-order user influence reflected in the user-user graph for user embedding learning. (DiffNet++通过将社会推荐转化为一个以社会网络和兴趣网络为输入的异构图,将反映在user-item图中的高阶用户潜在兴趣和反映在user-user图中的高阶用户影响注入DiffNet,以进行用户嵌入学习。)
- (3)This is achieved by iteratively aggregating each user’s embedding from three aspects: (这是通过从三个方面迭代聚合每个用户的嵌入来实现的:)
- the user’s previous embedding, (用户之前的嵌入,)
- the influence aggregation of social neighbors from the social network, (社交网络中社交邻居的影响力聚集)
- and the interest aggregation of item neighbors from the user-item interest network. (来自用户项目兴趣网络的项目邻居的兴趣聚合。)
- (4)Furthermore, we design a multi-level attention network that learns how to attentively aggregate user embeddings from these three aspects. (此外,我们设计了一个多层次的注意网络,学习如何从这三个方面区分注意力地集中用户嵌入)
recommender systems, graph neural network, social recommendation, influence diffusion, interest diffusion
(1) Collaborative Filtering (CF) based recommender systems learn user and item embeddings by utilizing user-item interest behavior data, and have attracted attention from both the academia and industry [37], [32]. However, as most users have limited behavior data, CF suffers from the data sparsity issue [1]. With the development of social networks, users build social relationships and share their item preferences on these platforms. As well supported by the social influence theory, users in a social network would influence each other, leading to similar preferences [12], [2]. Therefore, social recommendation has emerged, which focuses on exploiting social relations among users to alleviate data sparsity and enhancing recommendation performance [19], [20], [14], [43]. (基于协同过滤(CF)的推荐系统通过利用用户项目兴趣行为数据来学习用户和项目嵌入,已经引起了学术界和工业界的关注[37],[32]。然而,由于大多数用户的行为数据有限,CF存在数据稀疏问题[1]。随着社交网络的发展,用户在这些平台上建立社交关系并分享他们的商品偏好。社交网络中的用户会相互影响,从而产生相似的偏好[12],[2],这也得到了社会影响理论的支持。因此,社交推荐应运而生,其重点是利用用户之间的社交关系来缓解数据稀疏性,并提高推荐性能[19]、[20]、[14]、[43]。)
(2) In fact, as users play a central role in social platforms with user-user social behavior and user-item interest behavior, the key to social recommendation relies on learning user embeddings with these two kinds of behaviors. (事实上,由于用户在社交平台中扮演着核心角色,具有用户社交行为和用户项目兴趣行为,社交推荐的关键在于学习用户嵌入这两种行为。)
(3) Despite the performance improvement of these social recommendation models, we argue that the current social recommendation models are still far from satisfactory. (尽管这些社会推荐模型的性能有所提高,但我们认为目前的社会推荐模型仍然远远不能令人满意。)
In fact, as shown in Fig. 1, users play a central role in two kinds of behavior networks: the user-user social network and the user-item interest network. (事实上,如图1所示,用户在两种行为网络中扮演着核心角色:用户社交网络和用户项目兴趣网络。)
(4) To this end, we reformulate users’ two kinds of behaviors as a heterogeneous network with two graphs, i.e, a user-user social graph and a user-item interest graph, and propose how to explore the heterogeneous graph structure for social recommendation.
In fact, Graph Convolutional Networks (GCNs) have shown huge success for learning graph structures with theoretical elegance, practical flexibility and high performance [5], [8], [22]. GCNs perform node feature propagation in the graph, which recursively propagate node features by iteratively convolutional aggregations from neighborhood nodes, such that the up to K-th order graph structure is captured with K iterations [47]. (事实上,图卷积网络(GCN)在学习图结构方面取得了巨大的成功,具有理论上的优雅、实用的灵活性和高性能[5]、[8]、[22]。GCN在图中执行节点特征传播,通过迭代卷积聚合从邻域节点递归传播节点特征,从而通过K次迭代捕获高达K阶的图结构[47]。)
By treating user-item interactions as a bipartite interest graph and user-user social network as a social graph, some works have applied GCNs separately on these two kinds of graphs [51], [41], [48], [43]. (通过将用户项目交互视为一个二分兴趣图,将用户社交网络视为一个社交图,一些研究将GCN分别应用于这两类图[51]、[41]、[48]、[43]。)
(5) In this paper, we propose to advance our preliminary DiffNet structure, and jointly model the two graph structure (user-item graph and user-user graph) for social recommendation. (在本文中,我们提出了我们初步的DiffNet结构,并为社会推荐的两种图结构(用户项目图和用户用户图)联合建模。)
To this end, we propose DiffNet++, an improved algorithm of DiffNet that models the neural influence diffusion and interest diffusion in a unified framework. (为此,我们提出了DiffNet++,这是一种改进的DiffNet算法,它在一个统一的框架中对神经影响扩散和兴趣扩散进行建模。)
Furthermore, we design a multi-level attention network structure that learns how to attentively aggregate user embeddings from different nodes in a graph, and then from different graphs. (此外,我们设计了一个多层次的注意网络结构,学习如何专注地聚合来自图中不同节点的用户嵌入,然后再从不同的图中聚合。)
(6) In summary, our main contributions are listed as follows:
Extensive experimental results on two real-world datasets clearly show the effectiveness of our proposed DiffNet++ model. Compared to the baseline with the best performance, DiffNet++ outperforms it (在两个真实数据集上的大量实验结果清楚地表明了我们提出的DiffNet++模型的有效性。与性能最好的基线相比,DiffNet++的性能优于它)
(1) In a social recommender system, there are two sets of entities: a user set U U U (|U|=M), and an item set V V V (|V |=N). Users form two kinds of behaviors in the social platforms: making social connections with other users and showing item interests.
(2) Given the two kinds of users’ behaviors, the user-user social network is denoted as a user-user directed graph: G S = < U , S > GS=< U,S > GS=<U,S>,
The user interest network denotes users’ interests for items, which could be constructed from the user-item rating matrix R R R as an undirected bipartite network: G I = < U ∪ V , R > G_I = < U \cup V, R > GI=<U∪V,R>.
(3) Besides, each user a a a is associated with real-valued attributes (e.g, user profile), denoted as x a x_a xa in the user attribute matrix X ∈ R d 1 × M X \in R^{d1\times M} X∈Rd1×M.
In this subsection, we summarize the related works for social recommendation into three categories: classical social recommendation models, the recent graph based recommendation models, and attention modeling in the recommendation domain. (在这一小节中,我们将社会推荐的相关工作归纳为三类:经典的社会推荐模型、最新的基于图的推荐模型和推荐领域中的注意建模。)
(1) By formulating users’ historical behavior as a user-item interaction matrix R, most classical CF models embed both users and items in a low dimension latent space, such that each user’s predicted preference to an unknown item turns to the inner product between the corresponding user and item embeddings as [37], [32], [36]: (通过将用户的历史行为表述为用户-项目交互矩阵R,大多数经典CF模型将用户和项目嵌入到低维潜在空间中,使得每个用户对未知项目的预测偏好转向相应用户和项目嵌入之间的内积[37]、[32]、[36]:)
(2) In fact, as various specialized matrix factorization models have been proposed for specific tasks, factorization machines is proposed as a general approach to mimic most factorization models with simple feature engineering [36]. Recently, some deep learning based models have been proposed to tackle the CF problem [17], [28]. These approaches advanced previous works by modeling the non-linear complex interactions between users, or the complex interactions between sparse feature input. (事实上,由于针对特定任务提出了各种专门的矩阵分解模型,分解机被提议作为一种通用方法,用简单的特征工程模拟大多数分解模型[36]。最近,有人提出了一些基于深度学习的模型来解决CF问题[17],[28]。这些方法通过建模用户之间的非线性复杂交互,或稀疏特征输入之间的复杂交互,推进了以前的工作。)
(3) The social influence and social correlation among users’ interests are the foundation for building social recommender systems[29], [38], [25], [24]. Therefore, the social network among users could be leveraged to alleviate the sparsity in CF and enhance recommendation performance [30], [38], [15]. Due to the superiority of embedding based models for recommendation, most social recommendation models are also built on these embedding models. (用户兴趣的社会影响和社会关联是构建社会推荐系统的基础[29 ]、[38 ]、[25 ]、[24 ]。因此,可以利用用户之间的社交网络来缓解CF中的稀疏性,并提高推荐性能[30]、[38]、[15]。由于基于嵌入的推荐模型的优越性,大多数社会推荐模型也建立在这些嵌入模型的基础上。)
These social embedding models could be summarized into the following two categories: (这些社会嵌入模型可归纳为以下两类:)
(4) Instead of the social regularization term, some researchers argued that the social network provides valuable information to enhance each user’s behavior [50], [15]. TrustSVD is such a representative model that shows state-of-the-art performance [14], [15]. By assuming the implicit feedbacks of a user’s social neighbors’ on items could be regarded as the auxiliary feedback of this user, TrustSVD modeled the predicted preference as: (一些研究人员认为,社交网络提供了有价值的信息来增强每个用户的行为,而不是社交规范化术语[50],[15]。TrustSVD就是这样一个具有代表性的模型,它展示了最先进的性能[14],[15]。通过假设用户的社交邻居对项目的隐性反馈可以被视为该用户的辅助反馈,TrustSVD将预测偏好建模为:)
(5) As items are associated with attribute information (e.g., item description, item visual information), ContextMF is proposed to combine social context and social network under a collective matrix factorization framework with carefully designed regularization terms [21]. (由于项目与属性信息(如项目描述、项目视觉信息)相关联,ContextMF被提议在一个集体矩阵分解框架下结合社会背景和社会网络,并使用精心设计的正则化术语[21]。)
Social recommendation has also been extended with social circles [34], temporal context [38], rich contextual information [42], user role in the social network [39], and efficient training models without negative sampling [6]. All these previous works focused on how to explore the social neighbors, i.e., the observed links in the social network. (社交圈[34]、时间上下文[38]、丰富的上下文信息[42]、社交网络中的用户角色[39]以及无负采样的高效培训模型[6]也扩展了社交推荐。所有这些之前的工作都集中在如何探索社交邻居,即社交网络中观察到的链接。)
Recently, CNSR is proposed to leverage the global social network in the recommendation process [44]. In CNSR, each user’s latent embedding is composed of two parts: (最近,CNSR被提议在推荐过程中利用全球社交网络[44]。在CNSR中,每个用户的潜在嵌入由两部分组成:)
(1) GCNs generalize the convolutional operations from the regular Euclidean domains to non-Euclidean graph and have empirically shown great success in graph representation learning [5], [8], [22]. (GCN将卷积运算从正则欧几里德域推广到非欧几里德图,并在图表示学习[5]、[8]、[22]方面取得了巨大的成功。)
Specifically, GCNs recursively perform message passing by applying convolutional operations to aggregate the neighborhood information, such that the K-th order graph structure is captured with K iterations [22]. By treating the user-item interaction as a graph structure, GCNs have been applied for recommendation [48], [51]. (具体而言,GCN通过应用卷积运算来聚合邻域信息,从而递归地执行消息传递,从而通过K次迭代捕获K阶图结构[22]。通过将用户项目交互视为一个图形结构,GCN已被应用于建议[48],[51]。)
Earlier works relied on spectral GCNs, and suffered from huge time complexity [33], [51]. Therefore, many recent works focus on the spatial based GCNs for recommendation [48], [4], [49], [41]. (早期的工作依赖于光谱GCN,并且具有巨大的时间复杂性[33],[51]。因此,最近的许多工作都集中在基于空间的GCN推荐上)
(2) As the social structure among users could be naturally formulated as a user-user graph, recently we propose a preliminary graph based social recommendation model, DiffNet, for modeling the social diffusion process in recommendation [43]. DiffNet advances classical embedding based models with carefully designed influence diffusion layers, such that how users are influenced by the recursive influence diffusion process in the social network could be well modeled. (由于用户之间的社会结构可以自然地表示为用户-用户图,最近我们提出了一个初步的基于图的社会推荐模型DiffNet,用于在推荐中建模社会扩散过程[43]。DiffNet提出了经典的基于嵌入的模型,并精心设计了影响扩散层,从而可以很好地模拟社交网络中递归影响扩散过程对用户的影响。)
(3) In summary , all these previous GCN based models either considered the higher-order social network or the higher-order user interest network for recommendation. (综上所述,之前所有这些基于GCN的模型要么考虑高阶社交网络,要么考虑高阶用户兴趣网络进行推荐。)
There are some recently works that also leverage the graph neural networks for social recommendation [10], [45]. Specifically, (最近有一些研究也利用图形神经网络进行社会推荐[10],[45]。明确地)
(1) For each user a a a, the fusion layer takes p a p_a pa and her associated feature vector x a x_a xa as input, and outputs a user fusion embedding u a 0 u^0_a ua0 that captures the user’s initial interests from different kinds of input data. We model the fusion layer as: (对于每个用户 a a a,融合层将 p a p_a pa及其关联的特征向量 x a x_a xa作为输入,并输出用户融合嵌入 u a 0 u^0_a ua0,从不同类型的输入数据中捕获用户的初始兴趣。我们将融合层建模为:)
(2) Similarly , for each item i i i, the fusion layer models the item embedding v i 0 v^0_i vi0 as a function between its free latent vector q i q_i qi and its feature vector y i y_i yi as: (类似地,对于每个项目 i i i,融合层对嵌入 v i 0 v^0_i vi0的项目进行建模作为自由潜向量 q i q_i qi及其特征向量 y i y_i yi之间的函数)
(1) By feeding the output of each user a’s fused embedding u a 0 u^0_a ua0 and each item i i i’s fused embedding v i 0 v^0_i vi0 into the influence and interest diffusion layers, these layers recursively model the dynamics of this user’s latent preference and the item’s latent preference propagation in the graph G G G with layer-wise convolutions. (通过输入每个用户 a a a的融合嵌入 u a 0 u^0_a ua0的输出以及每一项 i i i的融合嵌入 v i 0 v^0_i vi0在影响层和兴趣扩散层中,在图 G G G中这些层使用分层卷积递归地建模该用户的潜在偏好和项目的潜在偏好传播的动力学。)
(2) For each item i i i, given its k k k-th layer embedding v i k v^k_i vik, we model the updated item embedding v i k + 1 v^{k+1}_i vik+1 at the ( k + 1 ) (k+1) (k+1)-th layer from G I G_I GI as:
(3) In the item neighbor aggregation function, Eq.(8) shows the weight of user a a a to item i i i. A naive idea is to aggregate the embeddings from i i i’s neighbor users with mean pooling operation, i.e., v ^ i k + 1 = ∑ a ∈ R i 1 ∣ R i ∣ u a k \hat{v}^{k+1}_i = \sum_{a\in R_i} \frac{1} {|Ri|} u^k_a v^ik+1=∑a∈Ri∣Ri∣1uak. However, it neglects the different interest weights from users, as the importance values of different users vary in item representation. Therefore, we use an attention network to learn the attentive weight η i a k + 1 \eta^{k+1}_{ia} ηiak+1 in Eq.(8) as: (然而,由于不同用户在项目表示中的重要性值不同,它忽略了用户的不同兴趣权重。因此,我们使用注意网络来学习注意权重 η i a k + 1 \eta^{k+1}_{ia} ηiak+1在等式(8)中,如下所示:)
(4) For each user a a a, let u a k u^k_a uak denote her latent embedding at the k k k-th layer. As users play a central role in both the social network GSand the interest network G I G_I GI, besides her own latent embedding u a k u^k_a uak, her updated embedding u a ( k + 1 ) u^{(k+1)}_a ua(k+1) at ( k + 1 ) (k + 1) (k+1)-th layer is influenced by two graphs:
(5) In addition to the user and item embeddings, there are three groups of weights in the above three equations. A naive idea is to directly et equal values of each kind of weights, i.e., γ a 1 ( k + 1 ) = γ a 2 ( k + 1 ) = 1 2 , α a b ( k + 1 ) = 1 ∣ S a ∣ \gamma^{(k+1)}_{a1} = \gamma^{(k+1)}_{a2} = \frac{1}{2}, \alpha^{(k+1)}_{ab} = \frac{1}{|Sa|} γa1(k+1)=γa2(k+1)=21,αab(k+1)=∣Sa∣1, and β ( k + 1 ) a i = 1 ∣ R a ∣ β(k+1) ai = 1 |Ra| β(k+1)ai=1∣Ra∣. (除了用户和项目嵌入,上述三个等式中还有三组权重。一个天真的想法是直接将每种权重的值相等)
However, this simple idea could not well model the different kinds of weights in the user decision process. In fact, these three groups of weights naturally present a two-layer multi-level structure. (然而,这种简单的想法无法很好地模拟用户决策过程中的各种权重。事实上,这三组权重自然呈现出两层多层次的结构。)
Specifically, the social influence strengths and the interest strengths could be seen as node-level weights, which model how each user balances different neighboring nodes in each graph.
By sending the aggregations of node level attention into Eq.(12), γ a l k + 1 \gamma^{k+1}_{al} γalk+1 is the graph level weight that learns to fuse and aggregate information from different graphs. Specifically, the graph layer weights are important as they model how each user balances the social influences and her historical records for user embedding. Different users vary, with some users are more likely to be swayed by the social network while the interests of others are quite stable. Therefore, the weights in the graph attention layer for each user also need to be personally adapted. (通过将节点级注意力的聚集发送到等式(12), γ a l k + 1 \gamma ^{k+1}_{al} γalk+1是学习融合和聚合来自不同图形的信息的图形级权重。具体来说,图层权重很重要,因为它们模拟了每个用户如何平衡社会影响和用户嵌入的历史记录。不同的用户各不相同,一些用户更容易受到社交网络的影响,而另一些用户的兴趣则相当稳定。因此,每个用户在图形注意层中的权重也需要进行个人调整。)
(6) As the three groups of weights represent a multi-level structure, we therefore use a multi-level attention network to model the attentive weights. (由于这三组权重代表一个多层次的结构,因此我们使用一个多层次的注意网络来建模注意权重。)
Specifically, the graph attention network is designed to learn the contribution weight of each aspect when updating a’s embedding with different graphs, i.e., p ^ a k + 1 \hat{p}^{k+1}_a p^ak+1 and q ^ a k + 1 \hat{q}^{k+1}_a q^ak+1 in Eq.(12), (具体来说,图注意网络的设计是为了在用不同的图更新a的嵌入时,了解每个方面的贡献权重,)
and the node attention networks are designed to learn the attentive weights in each social graph and each interest graph respectively . Specifically, the social influence score α a b k + 1 α^{k+1}_{ab} αabk+1 is calculated as follows: (节点注意网络分别用于学习每个社会图和每个兴趣图中的注意权重。具体来说,社会影响评分 α a b k + 1 α^{k+1}_{ab} αabk+1计算方法如下)
(7) In the above equation, the social influence strength α a b k + 1 \alpha^{k+1}_{ab} αabk+1 takes the related two users’ embeddings at the k k k-th layer as input, and sending these features into a MLP to learn the complex relationship between features for social influence strength learning. Without confusion, we omit the normalization step of all attention modeling in the following, as all of them share the similar form as shown in Eq.(11). (在上述等式中,社会影响强度 α a b k + 1 \alpha^{k+1}_{ab} αabk+1将相关的两个用户在 k k k-th层的嵌入作为输入,并将这些特征发送到MLP中,以了解特征之间的复杂关系,从而进行社会影响强度学习。毫无疑问,我们在下文中省略了所有注意力建模的规范化步骤,因为它们都具有类似的形式,如等式(11)所示。)
(8) Similarly, we calculate the interest influence score β a i k + 1 \beta^{k+1}_{ai} βaik+1 by taking related user embedding and item embedding as input: (同样,我们计算利息影响分数 β a i k + 1 \beta^{k+1}_{ai} βaik+1以相关用户嵌入和项目嵌入为输入:)
(9) After obtaining the two groups of the node attentive weights, the output of the node attention weights are sent to the graph attention network, and we could model the graph attention weights of γ a l k + 1 \gamma^{k+1}_{al} γalk+1 ( l l l = 1,2) as: (在获得两组节点注意权值后,将节点注意权值的输出发送到图注意网络,我们可以对节点的图注意权值进行建模)
(10) In the above equation, for each user a a a, the graph attention layer scores not only rely on the user’s embedding ( u a k ) (u^k_a) (uak), but also the weighted representations that are learnt from the node attention network. (在上面的等式中,对于每个用户 a a a,图注意层分数不仅取决于用户的嵌入( u a k u^k_a uak), 还包括从节点注意网络中学习到的加权表示。)
(11) For example, as shown in Eq.(12), γ a 1 ( k + 1 ) \gamma^{(k+1)}_{a1} γa1(k+1) denotes the influence diffusion weight for contributing to users’ depth ( k + 1 ) (k + 1) (k+1) embedding , with additional input of the learned attentive combination of the influence diffusion aggregation in Eq.(13). (表示有助于用户深度 ( k + 1 ) (k+1) (k+1)嵌入的影响扩散权重,并额外输入公式(13)中学习到的影响扩散聚合的注意组合。)
(1) We use a pair-wise ranking based loss function for optimization, which is widely used for implicit feedback [37]: (我们使用基于成对排序的损失函数进行优化,该函数广泛用于隐式反馈[37]:)
(2) For all the trainable parameters, we initialize them with the Gaussian distribution with a mean value of 0 and a standard deviation of 0.01.
(1) The key idea of our proposed DiffNet++ model is the well designed interest and influence diffusion layers. In fact, this part could be calculated in matrix forms. In the following, we would like to show how to update user and item embedding from the k-th layer to the (k + 1)-th layer with matrix operations. Let H ( k + 1 ) = [ η i a k + 1 ] ∈ R N × M H^{(k+1)} = [\eta^{k+1}_{ia}] \in R^{N\times M} H(k+1)=[ηiak+1]∈RN×M denote the matrix representation of attentive item aggregation weigth in Eq.(10), we have: (我们提出的DiffNet++模型的关键思想是精心设计的兴趣和影响扩散层。事实上,这一部分可以用矩阵形式计算。在下面,我们将展示如何使用矩阵运算将用户和项目嵌入从第k层更新到(k+1)层。设 H ( k + 1 ) = [ η i a k + 1 ] ∈ R N × M H^{(k+1)} = [\eta^{k+1}_{ia}] \in R^{N\times M} H(k+1)=[ηiak+1]∈RN×M表示式(10)中注意事项聚合权重的矩阵表示,我们有:)
(2) At the user side, given Eq.(12) ,
(3) After learning the attention matrices, we could update user and item embeddings at the (k + 1)-th layer as: (在学习了注意矩阵之后,我们可以将(k+1)层的用户和项目嵌入更新为:)
(4) Based on the above matrix operations of the social and influence diffusion layers, DiffNet++ is easily implemented by current deep learning frameworks. (基于上述社交和影响扩散层的矩阵运算,DiffNet++很容易通过当前的深度学习框架实现。)
(1) The proposed DiffNet++ model is designed under the problem setting with the input of user feature matrix X X X, item feature matrix Y Y Y, and the social network S. (提出的DiffNet++模型是在问题设置下设计的,输入用户特征矩阵 X X X、项目特征矩阵 Y Y Y和社交网络 S S S。)
(2) When the user (item) features are not available, the fusion layer disappears. In other words, as shown in Eq.(7), each item’s latent embedding v i 0 v^0_i vi0 degenerates to q i q_i qi. (当用户(项目)特征不可用时,融合层消失。换句话说,如等式(7)所示,每个项目的潜在嵌入 v i 0 v^0_i vi0退化为 q i q_i qi .)
(1) We conduct experiments on four real-world datasets: Yelp, Flickr, Epinions and Dianping. (我们在四个真实数据集上进行了实验:Yelp、Flickr、Epinions和Dianping。)
(2) Yelp is a well-known online location based social network, where users could make friends with others and review restaurants. We use the Yelp dataset that is publicly available2. (Yelp是一个著名的基于位置的在线社交网络,用户可以在这里与他人交朋友并查看餐厅。我们使用公开的Yelp数据集2。)
(3) Among the four datasets, Yelp and Flickr are two datasets with user and item attributes, and are adopted as datasets of our previously proposed DiffNet model [43]. (在这四个数据集中,Yelp和Flickr是两个具有用户和项目属性的数据集,它们被用作我们之前提出的DiffNet模型的数据集[43]。)