Embedding cardinality constraints in neural link predictors

Neuro-Symbolic Artificial Intelligence (AI) focuses on integrating symbolic and sub-symbolic systems to enhance the performance and explainability of predictive models. Symbolic and sub-symbolic approaches differ fundamentally in how they represent data and make use of data features to reach conclusions. Neuro-symbolic systems have recently received significant attention in the scientific community. However, despite efforts in neural-symbolic integration, symbolic processing can still be better exploited, mainly when these hybrid approaches are defined on top of knowledge graphs. This work is built on the statement that knowledge graphs can naturally represent the convergence between data and their contextual meaning (i.e., knowledge). We propose a hybrid system that resorts to symbolic reasoning, expressed as a deductive database, to augment the contextual meaning of entities in a knowledge graph, thus, improving the performance of link prediction implemented using knowledge graph embedding (KGE) models. An entity context is defined as the ego network of the entity in a knowledge graph. Given a link prediction task, the proposed approach deduces new RDF triples in the ego networks of the entities corresponding to the heads and tails of the prediction task on the knowledge graph (KG). Since knowledge graphs may be incomplete and sparse, the facts deduced by the symbolic system not only reduce sparsity but also make explicit meaningful relations among the entities that compose an entity ego network. As a proof of concept, our approach is applied over a KG for lung cancer to predict treatment effectiveness. The empirical results put the deduction power of deductive databases into perspective. They indicate that making explicit deduced relationships in the ego networks empowers all the studied KGE models to generate more accurate links.

Knowledge graph embedding models are used to learn low‐dimensional representations of entities and relations in knowledge graphs. In this paper, we propose Multi‐RAttE, an attention‐based learning method for multiple relational knowledge graph embedding representation, which divides the information transfer in the knowledge graph into cross‐relational information transfer and relation‐specific information transfer, and divides the embedding of knowledge graph entities into structural embedding and multi‐relational embedding for joint learning. To objectively analyse the performance of the Multi‐RAttE model, we select two typical datasets and different representative baseline models for experimental evaluation on several tasks such as link prediction, multi‐relation prediction and node classification. The experimental results show that the Multi‐RAttE model improves 8% over the state‐of‐the‐art model Composition‐based Multi‐Relational Graph Convolutional Networks (CompGCN) in terms of Hits@1 metric on the link prediction task on FB15k‐237 dataset; on the multi‐relation prediction task, the accuracy improves by 1.8% and 3.7% in the auc metric and F1 metric, respectively. The experimental results have proved that the Multi‐RAttE model can effectively perform the representation of multiple relations.

In recent years, network embedding has attracted much attention from researchers and achieved excellent performance. But few works investigate the adaptability of network embedding, especially for performance in different network structures. Heterogeneity, as a universal topological characteristic, plays a prominent role in network behaviors. In this study, we investigate the effect of heterogeneity on the effectiveness of existing network embedding approaches. We conduct experiments in scale-free networks with varying power exponents from both macro and micro perspectives to address link prediction and node similarity tasks, respectively. The results indicate that network embedding approaches can be divided into two classes according to their performance in the link prediction task. As the network heterogeneity decreases, the performance of approaches in the first class declines, while the performance of approaches in the second class initially improves and then declines. Moreover, our simulation discovers that, based on the node similarity metric, nodes are partitioned into two clusters by approaches, corresponding to large-degree nodes and small-degree nodes, respectively. Furthermore, approaches in the same class present similar characteristics between large-degree nodes and small-degree nodes, and the embedding is interpreted to some extent. Performance variations in the link prediction task can be explained by the characteristics of approaches, and similar characteristics are confirmed in experiments on real networks. Based on the findings for link prediction, we offer a brief guide for choosing an appropriate method based on the extent of heterogeneity. The investigation provides insight into network embedding and offers some interpretation of embedding.

Knowledge graphs (KGs) are widely used in many real-world applications, such as information retrieval, question answering system, and personal recommendation. However, most KGs are suffering from the incompleteness problem. To deal with the task of link prediction, previous knowledge graph embedding methods require numerous reference instances for each relation. It is worth noting that most relations in KGs have only a few reference instances available. Existing works for few-shot link prediction evaluate the authenticity of triplets from a single relation perspective. In this paper, we propose Hybrid Matching Network (HMNet) for few-shot link prediction, evaluating triplets from entity and relation two perspectives. At the entity-aware matching network, HMNet uses attentive inductive embedding layer to aggregate entity features and relation-aware topology, and then provides entity-aware score to implement first perspective evaluation. At the relation-aware matching network, HMNet integrates feature attention mechanism to implement relation perspective evaluation. Experiments on two public datasets indicate that HMNet achieves promising performance in few-shot link prediction.

Learning effective representations of entities and relations for knowledge graphs (KGs) is critical to the success of many multi-relational learning tasks. Existing methods based on graph neural networks learn a deterministic embedding function, which lacks sufficient flexibility to explore better choices when dealing with the imperfect and noisy KGs such as the scarce labeled nodes and noisy graph structure. To this end, we propose a novel multi-relational graph Gaussian Process network (GGPN), which aims to improve the flexibility of deterministic methods by simultaneously learning a family of embedding functions, i.e., a stochastic embedding function. Specifically, a Bayesian Gaussian Process (GP) is proposed to model the distribution of this stochastic function and the resulting representations are obtained by aggregating stochastic function values, i.e., messages, from neighboring entities. The two problems incurred when leveraging GP in GGPN are the proper choice of kernel function and the cubic computational complexity. To address the first problem, we further propose a novel kernel function that can explicitly take the diverse relations between each pair of entities into account and be adaptively learned in a data-driven way. We address the second problem by reformulating GP as a Bayesian linear model, resulting in a linear computational complexity. With these two solutions, our GGPN can be efficiently trained in an end-to-end manner. We evaluate our GGPN in link prediction and entity classification tasks, and the experimental results demonstrate the superiority of our method. Our code is available at https://github.com/sysu-gzchen/GGPN.

Network representation learning algorithms, which aim at automatically encoding graphs into low-dimensional vector representations with a variety of node similarity definitions, have a wide range of downstream applications. Most existing methods either have low accuracies in downstream tasks or a very limited application field, such as article classification in citation networks. In this paper, we propose a novel network representation method, named Link Prediction based Network Representation (LPNR), which generalizes the latest graph neural network and optimizes a carefully designed objective function that preserves linkage structures. LPNR can not only learn meaningful node representations that achieve competitive accuracy in node centrality measurement and community detection but also achieve high accuracy in the link prediction task. Experiments prove the effectiveness of LPNR on three real-world networks. With the mini-batch and fixed sampling strategy, LPNR can learn the embedding of large graphs in a few hours.

Knowledge Graph Completion (KGC) is a popular topic in knowledge graph construction and related applications, aiming to complete the structure of knowledge graphs by predicting missing entities or relations and mining unknown facts in the knowledge graph. In the KGC task, Graph Neural Network (GNN)-based methods have achieved remarkable results due to their advantage of effectively capturing complex relations among entities and generating more accurate and rich entity representations by aggregating information from neighbouring nodes. These methods mainly focus on the representation of entities, and the representation of relations is obtained using simple dimensional transformations or initial embeddings. This treatment ignores the diversity and complex semantics of relations and restricts the efficiency of the model in utilizing relational information in the reasoning process. In this work, we propose the Relational Representation Augmented Graph Attention Network (RRA-GAT), which effectively identifies and weights neighbouring relations that actually contribute to the target relation by filtering out irrelevant information through an attention function based on the information and spatial domain. Furthermore, we capture complex patterns and features in the relational embedding by means of a feed-forward network consisting of a series of linear transformations and nonlinear activation functions. Experiments demonstrate the very advanced performance of RRA-GAT on the link prediction task on standard datasets FB15k-237 and WN18RR (e.g., improved the MRR metric on the WN18RR dataset by 7.8% relative improvement).

Learning structured embeddings of knowledge graphs with generative adversarial framework

Multi-modal knowledge graphs representation learning via multi-headed self-attention

Knowledge graphs exhibit a typical hierarchical structure and find extensive applications in various artificial intelligence domains. However, large-scale knowledge graphs need to be completed, which limits the performance of knowledge graphs in downstream tasks. Knowledge graph embedding methods have emerged as a primary solution to enhance knowledge graph completeness. These methods aim to represent entities and relations as low-dimensional vectors, focusing on handling relation patterns and multi-relation types. Researchers need to pay more attention to the crucial feature of hierarchical relationships in real-world knowledge graphs. We propose a novel knowledge graph embedding model called Hierarchy-Aware Paired Relation Vectors Knowledge Graph Embedding (HPRE) to bridge this gap. By leveraging the power of 2D coordinates, HPRE adeptly model relation patterns, multi-relation types, and hierarchical features in the knowledge graph. Specifically, HPRE employs paired relation vectors to capture the distinct characteristics of head and tail entities, facilitating a better fit for relational patterns and multi-relation scenarios. Additionally, HPRE employs angular coordinates to differentiate entities at various levels of the hierarchy, effectively representing the hierarchical nature of the knowledge graph. The experimental results show that the HPRE model can effectively learn the hierarchical features of the knowledge graph and achieve state-of-the-art experimental results on multiple real-world datasets for the link prediction task.

The task of link prediction for knowledge graphs is to predict missing relationships between entities. Knowledge graph embedding, which aims to represent entities and relations of a knowledge graph as low dimensional vectors in a continuous vector space, has achieved promising predictive performance. If an embedding model can cover different types of connectivity patterns and mapping properties of relations as many as possible, it will potentially bring more benefits for link prediction tasks. In this paper, we propose a novel embedding model, namely LineaRE, which is capable of modeling four connectivity patterns (i.e., symmetry, antisymmetry, inversion, and composition) and four mapping properties (i.e., one-to-one, one-to-many, many-to-one, and many-to-many) of relations. Specifically, we regard knowledge graph embedding as a simple linear regression task, where a relation is modeled as a linear function of two low-dimensional vector-presented entities with two weight vectors and a bias vector. Since the vectors are defined in a real number space and the scoring function of the model is linear, our model is simple and scalable to large knowledge graphs. Experimental results on multiple widely used real-world datasets show that the proposed LineaRE model significantly outperforms existing state-of-the-art models for link prediction tasks.

Recently, a significant number of studies have focused on knowledge graph completion using rule-enhanced learning techniques, supported by the mined soft rules in addition to the hard logic rules. However, due to the difficulty in determining the confidences of the soft rules without the global semantics of knowledge graph such as the semantic relatedness between relations, the knowledge representation may not be optimal, leading to degraded effectiveness in its application to knowledge graph completion tasks. To address this challenge, this paper proposes a retrofit framework that iteratively enhances the knowledge representation and confidences of soft rules. Specifically, the soft rules guide the learning of knowledge representation, and the representation, in turn, provides global semantic of the knowledge graph to optimize the confidences of soft rules. Extensive evaluation shows that our method achieves new state-of-the-art results on link prediction and triple classification tasks, brought by the fine-tuned confidences of soft rules.

Embedding models have been successfully exploited for Knowledge Graph refinement. In these models, the data graph is projected into a low-dimensional space, in which graph structural information are preserved as much as possible, enabling an efficient computation of solutions. We propose a solution for injecting available background knowledge (schema axioms) to further improve the quality of the embeddings. The method has been applied to enhance existing models to produce embeddings that can encode knowledge that is not merely observed but rather derived by reasoning on the available axioms. An experimental evaluation on link prediction and triple classification tasks proves the improvement yielded implementing the proposed method over the original ones.

Derived from knowledge bases, knowledge graphs represent knowledge expressions in graphs, which utilize nodes and edges to denote entities and relations conceptually. Knowledge graph can be described in textual triple form, consisting of head entities, tail entities and relations between entities. In order to represent elements in knowledge graphs, knowledge graph embedding techniques are proposed to map entities and relations into continuous vector spaces as numeric vectors for computational efficiency. Convolution-based knowledge graph embedding models have promising performance for knowledge graph representation learning. However, the input of those neural network-based models is frequently in handmade forms and may suffer from low efficiency in feature extraction procedure of the models. In this paper, a convolutional autoencoder is proposed for knowledge graph representation learning with entity pairs as input, aiming to obtain corresponding hidden relation representation. In addition, a bi-directional relation encoding network is utilized to represent semantic of entities in different directional relation patterns, as an encoder to output representation for initialization of the convolutional autoencoder. Experiments are conducted on standard datasets including, WN18RR, Kinship, NELL-995 and FB15k-237 as a link prediction task. Besides, input embedding matrix composed of different ingredients is designed to evaluate performances of the convolutional autoencoder. The results demonstrate that our model is effective in learning representation from entity feature interactions.

Scoring functions (SFs), which measure the plausibility of triplets in knowledge graph (KG), have become the crux of KG embedding. Lots of SFs, which target at capturing different kinds of relations in KGs, have been designed by humans in recent years. However, as relations can exhibit complex patterns that are hard to infer before training, none of them can consistently perform better than others on existing benchmark data sets. In this paper, inspired by the recent success of automated machine learning (AutoML), we propose to automatically design SFs (AutoSF) for distinct KGs by the AutoML techniques. However, it is non-trivial to explore domain-specific information here to make AutoSF efficient and effective. We firstly identify a unified representation over popularly used SFs, which helps to set up a search space for AutoSF. Then, we propose a greedy algorithm to search in such a space efficiently. The algorithm is further sped up by a filter and a predictor, which can avoid repeatedly training SFs with same expressive ability and help removing bad candidates during the search before model training. Finally, we perform extensive experiments on benchmark data sets. Results on link prediction and triplets classification show that the searched SFs by AutoSF, are KG dependent, new to the literature, and outperform the state-of- the-art SFs designed by humans.