Relational Learning with Gated and Attentive Neighbor Aggregator for Few-Shot Knowledge Graph Completion

Niu, Guanglin; Li, Yang; Tang, Chengguang; Geng, Ruiying; Dai, Jian; Qiao, Li; Wang, Hao; Sun, Jian; Huang, Fei; Si, Luo

doi:10.1145/3404835.3462925

Cited by 43 publications

(36 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We consider five static KG FSL methods, i.e., Gmatching (Xiong et al, 2018), MetaR (Chen et al, 2019), FSRL (Zhang et al, 2020), FAAN (Sheng et al, 2020), GANA (Niu et al, 2021), and a TKG FSL method, i.e., OAT (Mirtaheri et al, 2021). In (Mirtaheri et al, 2021), static KG FSL methods are trained and evaluated on an unweighted static KG derived from collapsing the original TKG, which greatly decreases the inductive bias brougt by the original TKG and causes poor performance of these methods.…”

Section: Few-shot Relational Learning Methodsmentioning

confidence: 99%

“…Xiong et al (Xiong et al, 2018) find that data scarcity problem exists in KGs and traditional KG representation learning methods fail to model sparse KG relations. To solve this problem, several researches (Xiong et al, 2018;Chen et al, 2019;Zhang et al, 2020;Sheng et al, 2020;Niu et al, 2021) employ FSL paradigm (Vinyals et al, 2016;Sung et al, 2018) and make improvement in learning sparse relations. Mirtaheri et al (Mirtaheri et al, 2021) find the same problem in TKGs and develop a one-shot TKG reasoning model aiming to better model sparse relations in TKGs.…”

Section: Related Workmentioning

confidence: 99%

“…y ec q is the binary label indicating whether a candidate entity e c is the ground truth missing entity from q or not. Note that in Equation 6, for every query quadruple q, we average the loss over all the entities e c ∈ E other than o q (or s q ), rather than sampling a fixed number of negative samples as in previous KG FSL methods (Xiong et al, 2018;Chen et al, 2019;Zhang et al, 2020;Niu et al, 2021;Mirtaheri et al, 2021).…”

Section: Parameter Learningmentioning

confidence: 99%

See 2 more Smart Citations

Learning Meta Representations of One-shot Relations for Temporal Knowledge Graph Link Prediction

Ding¹,

He²,

Ma³

et al. 2022

Preprint

View full text Add to dashboard Cite

Few-shot relational learning for static knowledge graphs (KGs) has drawn greater interest in recent years, while few-shot learning for temporal knowledge graphs (TKGs) has hardly been studied. Compared to KGs, TKGs contain rich temporal information, thus requiring temporal reasoning techniques for modeling. This poses a greater challenge in learning few-shot relations in the temporal context. In this paper, we revisit the previous work related to few-shot relational learning in KGs and extend two existing TKG reasoning tasks, i.e., interpolated and extrapolated link prediction, to the one-shot setting. We propose four new large-scale benchmark datasets and develop a TKG reasoning model for learning one-shot relations in TKGs. Experimental results show that our model can achieve superior performance on all datasets in both interpolation and extrapolation tasks.

show abstract

Section: Few-shot Relational Learning Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Parameter Learningmentioning

confidence: 99%

See 1 more Smart Citation

Learning Meta Representations of One-shot Relations for Temporal Knowledge Graph Link Prediction

Ding¹,

He²,

Ma³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…To this end, it utilizes a relation-meta learner to generate relation information from the embeddings of entity pairs. GANA-MTransH [10] shares a similar idea with MetaR, but learns the relation-specific hyper-plane parameters to model complex relations. (3) Dual-process theory-based methods: Inspired by the cognitive system of human beings, Du et al [28] recently developed a Cognitive KG Reasoning model named CogKR, which collects information about the entities in the reference triples and then conduct relation reasoning over the collected information.…”

Section: B Few-shot Knowledge Graph Completionmentioning

confidence: 99%

“…Over the past decade, knowledge graph embedding (KGE) has proven to be a powerful technique for the KG completion task [6]- [8]. It represents each entity and each relation in KGs as a low-dimensional vector, and the plausibility of a triple can be measured based on its semantic representation [9], [10]. In particular, these KGE models all require sufficient training triples for both entities and relations.…”

Section: Introductionmentioning

confidence: 99%

Learning Relation-Specific Representations for Few-shot Knowledge Graph Completion

Li¹,

Yu²,

Zhang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Recent years have witnessed increasing interest in few-shot knowledge graph completion (FKGC), which aims to infer unseen query triples for a few-shot relation using a handful of reference triples of the relation. The primary focus of existing FKGC methods lies in learning the relation representations that can reflect the common information shared by the query and reference triples. To this end, these methods learn the embeddings of entities with their direct neighbors, and use the concatenation of the entity embeddings as the relation representations. However, the entity embeddings learned only from direct neighborhoods may have low expressiveness when the entity has sparse neighbors or shares a common local neighborhood with other entities. Moreover, the embeddings of two entities are insufficient to represent the semantic information of their relationship, especially when they have multiple relations. To address these issues, we propose a Relation-Specific Context Learning (RSCL) framework, which exploits graph contexts of triples to capture the semantic information of relations and entities simultaneously. Specifically, we first extract graph contexts for each triple, which can provide long-term entity-relation dependencies. To model the graph contexts, we then develop a hierarchical relation-specific learner to learn global and local relation-specific representations for relations by capturing contextualized information of triples and incorporating local information of entities. Finally, we utilize the learned representations to predict the likelihood of the query triples. Experimental results on two public datasets demonstrate that RSCL outperforms state-of-the-art FKGC methods.

show abstract

Tackling Solitary Entities for Few-Shot Knowledge Graph Completion

Liang

Zhao

Cheng

et al. 2022

Knowledge Science, Engineering and Management

View full text Add to dashboard Cite

Relational Learning with Gated and Attentive Neighbor Aggregator for Few-Shot Knowledge Graph Completion

Cited by 43 publications

References 32 publications

Learning Meta Representations of One-shot Relations for Temporal Knowledge Graph Link Prediction

Learning Meta Representations of One-shot Relations for Temporal Knowledge Graph Link Prediction

Learning Relation-Specific Representations for Few-shot Knowledge Graph Completion

Tackling Solitary Entities for Few-Shot Knowledge Graph Completion

Contact Info

Product

Resources

About