Multi-Task Feature Learning for Knowledge Graph Enhanced Recommendation

为学习翻译！！！

链接：https://scholar.google.com.hk/scholar?q=Multi-Task+Feature+Learning+for+Knowledge+Graph+Enhanced+Recommendation&hl=zh-CN&as_sdt=0&as_vis=1&oi=scholart

标题Multi-Task Feature Learning for Knowledge Graph Enhanced Recommendation

面向知识图增强推荐的多任务特征学习

摘要：
在实际的推荐场景中，协同过滤往往存在稀疏性和冷启动等问题，因此，研究人员和工程师通常使用侧信息来解决这些问题，提高推荐系统的性能。在本文中，我们将知识图作为边信息的来源。我们提出了一种多任务特征学习方法，用于知识图增强推荐。MKR是利用知识图嵌入任务辅助推荐任务的端到端深度框架。这两个任务通过交叉压缩单元相关联，该单元自动共享潜在特征，并学习推荐系统中的条目与知识图中的实体之间的高阶交互。我们证明了交叉压缩单元具有足够的多项式逼近能力，并证明了MKR是一个广义的框架，适用于几种有代表性的推荐系统和多任务学习方法。通过对真实世界数据集的广泛实验，我们证明了MKR在电影、书籍、音乐和新闻推荐方面取得了显著的进步，超过了最先进的基线。即使用户-项交互稀疏，MKR也能保持令人满意的性能。

Collaborative filtering often suffers from sparsity and cold start problems in real recommendation scenarios, therefore, researchers and engineers usually use side information to address the issues and improve the performance of recommender systems. In this paper, we consider knowledge graphs as the source of side information. We propose MKR, a Multi-task feature learning approach for Knowledge graph enhanced Recommendation. MKR is a deep end-to-end framework that utilizes knowledge graph embedding task to assist recommendation task. The two tasks are associated by cross&compress units, which automatically share latent features and learn high-order interactions between items in recommender systems and entities in the knowledge graph. We prove that cross&compress units have sufficient capability of polynomial approximation, and show that MKR is a generalized framework over several representative methods of recommender systems and multi-task learning. Through extensive experiments on real-world datasets, we demonstrate that MKR achieves substantial gains in movie, book, music, and news recommendation, over state-of-theart baselines. MKR is also shown to be able to maintain satisfactory performance even if user-item interactions are sparse.

介绍

推荐系统(RS)旨在解决信息爆炸，满足用户的个性化兴趣。最流行的推荐技术之一是协同过滤(CF)[11]，它利用用户的历史交互，并根据用户的共同偏好提供推荐。然而，基于cf的方法通常面临用户-物品交互的稀疏性和冷启动问题。因此，研究人员提出在推荐系统中使用边信息，包括社交网络[10]、属性[30]和多媒体(如文本[29]、图像[40])。知识图是知识管理的一种边信息，它通常包含丰富的事实和项目之间的联系。最近，研究人员提出了几个学术和商业KGs，如NELL1, DBpedia2，谷歌Knowledge Graph3和Microsoft Satori4。由于KG的高维性和异质性，通常采用知识图嵌入方法[27]对其进行预处理，在保持其固有结构的同时，将实体和关系嵌入到低维向量空间中。

Recommender systems (RS) aims to address the information explosion and meet users personalized interests. One of the most popular recommendation techniques is collaborative filtering (CF) [11], which utilizes users’ historical interactions and makes recommendations based on their common preferences. However, CF-based methods usually suffer from the sparsity of user-item interactions and the cold start problem. Therefore, researchers propose using side information in recommender systems, including social networks [10], attributes [30], and multimedia (e.g., texts [29], images [40]). Knowledge graphs (KGs) are one type of side information for RS, which usually contain fruitful facts and connections about items. Recently, researchers have proposed several academic and commercial KGs, such as NELL1, DBpedia2, Google Knowledge Graph3 and Microsoft Satori4. Due to its high dimensionality and heterogeneity, a KG is usually pre-processed by knowledge graph embedding (KGE) methods [27], which embeds entities and relations into low-dimensional vector spaces while preserving its inherent structure.

Existing KG-aware methods
受KG在各种任务中成功应用的启发，研究人员最近尝试利用KG来提高推荐系统的性能[31,32,39,40,45]。个性化实体推荐(Personalized Entity Recommendation, PER)[39]和分组拉索因子分解机(Factorization Machine with Group lasso, FMG)[45]将KG作为异构信息网络，提取基于元路径/元图的潜在特征，以表示不同类型关系路径/图上用户与项目之间的连接。值得注意的是，PER和FMG严重依赖于手动设计的元路径/元图，这限制了其在一般推荐场景中的应用。深度知识感知网络(Deep Knowledge-aware Network, DKN)[32]设计了一个CNN框架，将实体嵌入和词嵌入相结合进行新闻推荐。但是，在使用DKN之前需要进行实体嵌入，导致DKN缺乏端到端训练方式。DKN的另一个问题是，它几乎不能包含文本以外的侧面信息。RippleNet[31]是一个类似内存网络的模型，它在KG中传播用户的潜在偏好，并探索他们的层次兴趣。但是在RippleNet中关系的重要性是弱刻画的，因为一个关系R的嵌入矩阵很难被训练成二次形式v Rh (v和h是两个实体的嵌入向量)的重要性。协同知识库嵌入(Collaborative Knowledge base Embedding, CKE)[40]将CF与结构化知识、文本知识和可视化知识在一个统一的框架中结合起来。然而，CKE中的KGE模块(即TransR[13])更适合于图内应用(如KG完成和链接预测)而不是推荐。此外，CKE中CF模块和KGE模块在贝叶斯框架下松散耦合，使得KG对推荐系统的监督不那么明显。
Inspired by the success of applying KG in a wide variety of tasks, researchers have recently tried to utilize KG to improve the performance of recommender systems [31, 32, 39, 40, 45]. Personalized Entity Recommendation (PER) [39] and Factorization Machine with Group lasso (FMG) [45] treat KG as a heterogeneous information network, and extract meta-path/meta-graph based latent features to represent the connectivity between users and items along different types of relation paths/graphs. It should be noted that PER and FMG rely heavily on manually designed meta-paths/meta-graphs, which limits its application in generic recommendation scenarios. Deep Knowledge-aware Network (DKN) [32] designs a CNN framework to combine entity embeddings with word embeddings for news recommendation. However, the entity embeddings are required in advance of using DKN, causing DKN to lack an endto- end way of training. Another concern about DKN is that it can hardly incorporate side information other than texts. RippleNet [31] is a memory-network-like model that propagates users potential preferences in the KG and explores their hierarchical interests. But the importance of relations is weakly characterized in RippleNet, because the embedding matrix of a relation R can hardly be trained to capture the sense of importance in the quadratic form v Rh (v and h are embedding vectors of two entities). Collaborative Knowledge base Embedding (CKE) [40] combines CF with structural knowledge, textual knowledge, and visual knowledge in a unified framework. However, the KGE module in CKE (i.e., TransR [13]) is more suitable for in-graph applications (such as KG completion and link prediction) rather than recommendation. In addition, the CF module and the KGE module are loosely coupled in CKE under a Bayesian framework, making the supervision from KG less obvious for recommender systems.

The proposed approach

为了解决之前工作的局限性，我们提出了一种多任务学习(MTL)方法，用于知识图增强推荐。MKR是一个通用的端到端深度推荐框架，旨在利用KGE任务辅助推荐任务。注意，这两个任务不是相互独立的，但高度相关，因为RS中的一个项目可能与KG中的一个或多个实体相关。因此，一个项目及其对应实体在RS和KG中可能具有相似的邻近结构，在低层次和非任务特定的潜在特征空间[15]中可能具有相似的特征。我们将在实验部分进一步验证相似性。为了模拟物品和实体之间的共享特性，我们在MKR中设计了一个交叉压缩单元。交叉压缩单元明确地模拟了物品和实体特征之间的高阶交互，并自动控制了两种任务之间的交叉知识转移。通过交叉压缩单元，项目和实体的表示可以相互补充，帮助两项任务避免拟合噪声和提高泛化能力。通过对两个不同频率的任务交替优化，可以训练整个框架，使MKR在实际推荐场景中具有较高的灵活性和适应性。

To address the limitations of previous work, we propose MKR, a multi-task learning (MTL) approach for knowledge graph enhanced recommendation. MKR is a generic, end-to-end deep recommendation framework, which aims to utilize KGE task to assist recommendation task5. Note that the two tasks are not mutually independent, but are highly correlated since an item in RS may associate with one or more entities in KG. Therefore, an item and its corresponding entity are likely to have a similar proximity structure in RS and KG, and share similar features in low-level and non-task-specific latent feature spaces [15]. We will further validate the similarity in the experiments section. To model the shared features between items and entities, we design a cross&compress unit in MKR. The cross&compress unit explicitly models high-order interactions between item and entity features, and automatically control the cross knowledge transfer for both tasks. Through cross&compress units, representations of items and entities can complement each other, assisting both tasks in avoiding fitting noises and improving generalization. The whole framework can be trained by alternately optimizing the two tasks with different frequencies, which endows MKR with high flexibility and adaptability in real recommendation scenarios.

我们探讨了MKR的表达能力，并通过理论分析表明，交叉压缩单元能够接近项目和实体之间的高阶特征交互作用。我们还表明，MKR是一个广义框架，涵盖了推荐系统和多任务学习的几种代表性方法，包括分解机[22,23]、深度交叉网络[34]和十字绣网络[18]。根据经验，我们在四个推荐场景中评估我们的方法，即电影、书籍、音乐和新闻推荐。结果表明，在点击率(CTR)预测(例如，电影的平均AUC提高了11.6%)和top-K推荐(例如，书籍的平均Recall@10提高了66.4%)两方面，MKR都比最先进的基线取得了实质性的进展。即使在用户-项交互稀疏的情况下，MKR也能保持令人满意的性能。

We probe the expressive capability of MKR and show, through theoretical analysis, that the cross&compress unit is capable of approximating sufficiently high order feature interactions between items and entities. We also show that MKR is a generalized framework over several representative methods of recommender systems and multi-task learning, including factorization machines [22, 23], deep&cross network [34], and cross-stitch network [18]. Empirically, we evaluate our method in four recommendation scenarios, i.e., movie, book, music, and news recommendations. The results demonstrate that MKR achieves substantial gains over state-ofthe- art baselines in both click-through rate (CTR) prediction (e.g., 11.6% AUC improvements on average for movies) and top-K recommendation (e.g., 66.4% Recall@10 improvements on average for books). MKR is also shown to maintain satisfactory performance even when user-item interactions are sparse.

Contribution

值得注意的是，本文所研究的问题也可以建模为跨域推荐[26]或迁移学习[21]，因为我们更关心推荐任务的性能。然而，关键的观察是，虽然跨域推荐和迁移学习对目标域有单一的目标，但它们的损失函数仍然包含用于测量源域中数据分布或两个域之间相似性的约束项。在我们提出的MKR中，KGE任务显式地作为约束项，为推荐系统提供正则化。我们想强调的是，本文的主要贡献正是将问题建模为多任务学习：理论分析和实验结果表明，任务间相似度不仅对推荐系统有帮助，而且对知识图嵌入也有帮助，这比跨领域推荐和迁移学习更进一步。

It is worth noticing that the problem studied in this paper can also be modelled as cross-domain recommendation [26] or transfer learning [21], since we care more about the performance of recommendation task. However, the key observation is that though cross-domain recommendation and transfer learning have single objective for the target domain, their loss functions still contain constraint terms for measuring data distribution in the source domain or similarity between two domains. In our proposed MKR, the KGE task serves as the constraint term explicitly to provide regularization for recommender systems. We would like to emphasize that the major contribution of this paper is exactly modeling the problem as multi-task learning ：We go a step further than cross-domain recommendation and transfer learning by finding that the intertask similarity is helpful to not only recommender systems but also knowledge graph embedding, as shown in theoretical analysis and experiment results.

OUR APPROACH

在本节中，我们首先提出了知识图增强推荐问题，然后介绍了MKR的框架，并详细介绍了交叉压缩单元、推荐模块和KGE模块的设计。最后讨论了MKR的学习算法。

In this section, we first formulate the knowledge graph enhanced recommendation problem, then introduce the framework of MKR and present the design of the cross&compress unit, recommendation module and KGE module in detail. We lastly discuss the learning algorithm for MKR.

Problem Formulation
在本文中，我们制定了如下的知识图增强推荐问题。在一个典型的推荐场景中，我们有M个用户U = {u1,u2，…，uM}和N个元素V = {v1,v2，…vN}。根据用户隐式反馈定义用户-物品交互矩阵Y，其中 $y_{uv} = 1$ 表示用户参与item v，如点击、观看、浏览、购买等行为;否则 $y_{uv} = 0$ 。此外，我们还可以访问一个知识图G，它由实体-关系-实体三元组(h, r, t)组成。这里h、r和t分别表示知识三元组的头、关系和尾。例如，三元组(昆汀·塔伦蒂诺，电影导演。电影《低俗小说》中提到了昆汀·塔伦蒂诺执导的电影《低俗小说》。在许多推荐场景中，一个项目v可能与g中的一个或多个实体相关联。例如，在电影推荐中，“Pulp Fiction”这个项目在知识图中与其同名的实体相关联，而在新闻推荐中，标题为“特朗普在科技会议上承诺援助硅谷”的新闻，在知识图谱中与实体“唐纳德·特朗普”和“硅谷”联系在一起。

We formulate the knowledge graph enhanced recommendation problem in this paper as follows. In a typical recommendation scenario, we have a set of M users U = {u1,u2, … ,uM} and a set of N items V = {v1,v2, … ,vN }. The user-item interaction matrix Y RM N is defined according to users implicit feedback, where yuv = 1 indicates that useru engaged with itemv, such as behaviors of clicking, watching, browsing, or purchasing; otherwise yuv = 0. Additionally, we also have access to a knowledge graph G, which is comprised of entity-relation-entity triples (h, r , t ). Here h, r , and t denote the head, relation, and tail of a knowledge triple, respectively. For example, the triple (Quentin Tarantino, film.director. film, Pulp Fiction) states the fact that Quentin Tarantino directs the film Pulp Fiction. In many recommendation scenarios, an item v V may associate with one or more entities in G. For example, in movie recommendation, the item “Pulp Fiction” is linked with its namesake in a knowledge graph, while in news recommendation, news with the title “Trump pledges aid to Silicon Valley during tech meeting” is linked with entities “Donald Trump” and “Silicon Valley” in a knowledge graph.

在已知用户-物品交互矩阵Y和知识图G的情况下，我们的目的是预测用户u对之前没有交互的物品v是否有潜在的兴趣。我们的目标是学习一个预测函数y uv = F(u,v|Θ， y, G)，其中y uv表示用户u与物品v接触的概率，Θ是函数F的模型参数。

Given the user-item interaction matrix Y as well as the knowledge graph G, we aim to predict whether user u has potential interest in item v with which he has had no interaction before. Our goal is to learn a prediction function y uv = F(u,v|Θ, Y, G), where y uv denotes the probability that user u will engage with item v, and Θ is the model parameters of function F .

Framework

MKR的框架如图1a所示。MKR由三个主要部分组成:推荐模块、KGE模块和交叉压缩单元。(1)左边的推荐模块以用户和商品为输入，使用多层感知器(MLP)和交叉压缩单元，分别为用户和商品提取短特征和密集特征。然后将提取的特征一起输入另一个MLP，以输出预测的概率。(2)与左侧部分相似，右侧部分的KGE模块也使用多层提取知识三元组头部和关系的特征，并在得分函数f和真实尾部的监督下输出预测尾部的表示。(3)推荐模块与KGE模块之间采用专门设计的交叉压缩单元进行衔接。该单元能够自动学习推荐系统中条目与知识图中实体之间的高阶特征交互。

The framework of MKR is illustrated in Figure 1a. MKR consists of three main components: recommendation module, KGE module, and cross&compress units. (1) The recommendation module on the left takes a user and an item as input, and uses a multi-layer perceptron (MLP) and cross&compress units to extract short and dense features for the user and the item, respectively. The extracted features are then fed into another MLP together to output the predicted probability. (2) Similar to the left part, the KGE module in the right part also uses multiple layers to extract features from the head and relation of a knowledge triple, and outputs the representation of the predicted tail under the supervision of a score function f and the real tail. (3) The recommendation module and the KGE module are bridged by specially designed cross&compress units. The proposed unit can automatically learn high-order feature interactions of items in recommender systems and entities in the knowledge graph.

Cross&compress Unit

为了模拟物品与实体之间的交互，我们在MKR框架中设计了一个交叉压缩单元。如图1b所示，对于item v和它的一个关联实体e，我们首先从第l层构造它们的潜在特征vl和el的d*d成对交互作用:

To model feature interactions between items and entities, we design a cross&compress unit in MKR framework. As shown in Figure 1b, for itemv and one of its associated entities e, we first construct d d pairwise interactions of their latent feature vl Rd and el Rd from layer l :

然后通过将交叉特征矩阵投射到它们的潜在表示空间中，输出下一层的项目和实体的特征向量:

其中w·l Rd和b·l Rd是可训练的权重和偏倚向量。这被称为压缩操作，因为权重向量将交叉特征矩阵从Rd空间投射回特征空间Rd。注意，在Eq.(2)中，为了对称，交叉特征矩阵沿着水平和垂直方向被压缩(通过操作Cl和Cl)，但我们将在第3.2节提供更多的设计见解。为简单起见，交叉压缩单元记为

在接下来的文章中，我们使用后缀[v]或[e]来区分它的两种输出。通过交叉和压缩单元，MKR可以自适应地调整知识转移的权重，并学习两个任务之间的相关性。
需要注意的是，交叉压缩单元应该只存在于MKR的底层，如图1a所示。这是因为:(1)在深度架构中，**特征通常沿着网络由一般向特定转变，随着任务不相似度[38]的增加，特征在更高层次上的可转移性显著下降。**因此，共享高层可能存在负迁移的风险，尤其是对于MKR中的异构任务。(2)在MKR的高层中，项目特征与用户特征混合，实体特征与关系特征混合。混合特征不适合共享，因为它们没有明确的关联。

Recommendation Module

MKR中推荐模块的输入由两个原始特征向量u和v组成，分别描述用户u和物品v。u和v可以根据应用场景定制为one-hot ID[8]、属性[30]、词袋[29]或它们的组合。在给定用户u的原始特征向量u的情况下，我们使用L-layer MLP来提取其潜在的压缩特征:

The input of the recommendation module in MKR consists of two raw feature vectors u and v that describe user u and item v, respectively. u and v can be customized as one-hot ID [8], attributes [30], bag-of-words [29], or their combinations, based on the application scenario. Given user u s raw feature vector u, we use an L-layer MLP to extract his latent condensed feature:

在得到用户u的潜在特征uL和itemv的潜在特征vL后，我们通过预测函数fRS将两种路径结合起来，例如，内部产品或h层MLP。用户u参与道具v的最终预测概率为:

Knowledge Graph Embedding Module

知识图嵌入是将实体和关系嵌入到连续的向量空间中，同时保持其结构。近年来，研究人员提出了大量的KGE方法，包括翻译距离模型[2,13]和语义匹配模型[14,19]。在MKR中，我们提出了KGE模块的深度语义匹配体系结构。与推荐模块类似，对于给定的知识三元组(h, r, t)，我们首先利用多个交叉压缩单元和非线性层分别处理头h和关系r(包括ID[13]、类型[36]、文本描述[35]等)的原始特征向量。然后将它们的潜在特征连接在一起，然后用k层MLP预测尾部t:

Knowledge graph embedding is to embed entities and relations into continuous vector spaces while preserving their structure. Recently, researchers have proposed a great many KGE methods, including translational distance models [2, 13] and semantic matching models [14, 19]. In MKR, we propose a deep semantic matching architecture for KGE module. Similar to the recommendation module, for a given knowledge triple (h, r , t ), we first utilize multiple cross&compress units and nonlinear layers to process the raw feature vectors of head h and relation r (including ID [13], types [36], textual description [35], etc.), respectively. Their latent features are then concatenated together, followed by a K-layer MLP for predicting tail t :

其中S(h)为实体h的相关项集合，t为尾部t的预测向量。最后，使用评分(相似度)函数fKG计算三元组(h, r, t)的评分:

其中t是t的实特征向量。在本文中，我们使用归一化内积fKG(t, t) = σ(t t)作为评分函数[18]的选择，但其他形式的(dis)相似性度量也可以应用于此，如KullbackâĂŞLeibler发散。

Learning Algorithm

在Eq.(9)中，第一项度量的是推荐模块中的损失，其中u和v分别遍测用户集和项目集，J为交叉熵函数。第二项计算KGE模块中的损失，其中我们的目标是增加所有真三元组的得分，同时减少所有假三元组的得分。最后一项是防止过拟合的正则化项，λ1和λ2是平衡参数.

注意，Eq.(9)中的损失函数遍历所有可能的用户项对和知识三元组。为了提高计算效率，在[17]之后，我们在训练过程中使用了一个负采样策略。

MKR的学习算法在算法1中提出，其中训练历元包括两个阶段:推荐

3,THEORETICAL ANALYSIS

在这一节中，我们证明了交叉压缩单元具有足够的多项式逼近能力。我们还表明，MKR是一个广义的框架，涵盖了几种有代表性的推荐系统和多任务学习方法。

3.1 Polynomial Approximation
根据维尔斯特拉斯近似定理[25]，任何函数在一定的光滑性假设下都可以用多项式逼近到任意精度。因此，我们研究了交叉压缩单元的高阶相互作用近似的能力。我们证明了交叉压缩单元可以将物品-实体特征交互的顺序建模到指数级:

定理1指出，交叉和压缩单元可以自动地为足够高阶的项目和实体的组合特征建模，这证明了MKR的优越的近似能力与现有的工作，如广深[3]，分解机[22,23]和DCN[34]。

注意，定理1给出了交叉压缩单元的多项式逼近能力的理论观点，而不是提供其实际性能的保证。我们将在实验部分对交叉压缩装置进行实证评估。

3.2 Unified View of Representative Methods

在接下来的文章中，我们给出了推荐系统和多任务学习中几个有代表性的模型的统一观点，通过展示它们是MKR的限制版本或在理论上与MKR相关。这证明了交叉压缩单元的设计，并从概念上解释了与基线相比，其强大的经验性能。

3.2.1分解机器。分解机[22,23]是推荐系统的一种通用方法。在给定输入特征向量的情况下，FMs利用因式参数对输入向量中变量之间的所有交互作用进行建模，从而能够在推荐系统等稀疏性较大的问题中估计交互作用。二度因子分解机的模型方程定义为

命题1。v1和e1的l1范数可以写成如下形式

有趣的是注意到,而不是分解的重量参数xi xj成两个向量的点积在调频,词的重量vi ej映像成两个标量的和cross&压缩单位减少参数的数量和增加模型的鲁棒性。