Meta-Knowledge … (2021-10-27T00:00:00.000000Z)

TL;DR

A model MorsE is proposed, which does not learn embeddings for entities but learns transferable meta-knowledge that can be used to produce entity embeddINGS that significantly outperforms corresponding baselines for in-KGs and out-of-KG tasks in inductive settings.

Abstract

Knowledge graphs (KGs) consisting of a large number of triples have become widespread recently, and many knowledge graph embedding (KGE) methods are proposed to embed entities and relations of a KG into continuous vector spaces. Such embedding methods simplify the operations of conducting various in-KG tasks (e.g., link prediction) and out-of-KG tasks (e.g., question answering). They can be viewed as general solutions for representing KGs. However, existing KGE methods are not applicable to inductive settings, where a model trained on source KGs will be tested on target KGs with entities unseen during model training. Existing works focusing on KGs in inductive settings can only solve the inductive relation prediction task. They can not handle other out-of-KG tasks as general as KGE methods since they don't produce embeddings for entities. In this paper, to achieve inductive knowledge graph embedding, we propose a model MorsE, which does not learn embeddings for entities but learns transferable meta-knowledge that can be used to produce entity embeddings. Such meta-knowledge is modeled by entity-independent modules and learned by meta-learning. Experimental results show that our model significantly outperforms corresponding baselines for in-KG and out-of-KG tasks in inductive settings.

Authors

Mingyang Chen

4 papers

Hua-zeng Chen

10 papers

Wen Zhang

9 papers

References69 items

Meta-Learning Based Hyper-Relation Feature Modeling for Out-of-Knowledge-Base Embedding

AliCG: Fine-grained and Evolvable Conceptual Graph Construction for Semantic Search at Alibaba

Learning to Pre-train Graph Neural Networks

Meta-Inductive Node Classification across Graphs

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

Abstract

References69 items

TL;DR

Abstract

Authors

References69 items

Meta-Learning Based Hyper-Relation Feature Modeling for Out-of-Knowledge-Base Embedding

AliCG: Fine-grained and Evolvable Conceptual Graph Construction for Semantic Search at Alibaba

Learning to Pre-train Graph Neural Networks

Meta-Inductive Node Classification across Graphs

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

TL;DR

Abstract

Authors

References69 items

Meta-Learning Based Hyper-Relation Feature Modeling for Out-of-Knowledge-Base Embedding

AliCG: Fine-grained and Evolvable Conceptual Graph Construction for Semantic Search at Alibaba

Learning to Pre-train Graph Neural Networks

Meta-Inductive Node Classification across Graphs

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

QA-GNN: Reasoning with Language Models and Knowledge Graphs for Question Answering

Topology-Aware Correlations Between Relations for Inductive Link Prediction in Knowledge Graphs

Communicative Message Passing for Inductive Relation Reasoning

FedE: Embedding Knowledge Graphs in Federated Setting

Adaptive Attentional Network for Few-Shot Knowledge Graph Completion

DualDE: Dually Distilling Knowledge Graph Embedding for Faster and Cheaper Reasoning

GCC: Graph Contrastive Coding for Graph Neural Network Pre-Training

Learning to Extrapolate Knowledge: Transductive Few-shot Out-of-Graph Link Prediction

Scalable Multi-Hop Relational Reasoning for Knowledge-Aware Question Answering

Structure-Augmented Text Representation Learning for Efficient Knowledge Graph Completion

You CAN Teach an Old Dog New Tricks! On Training Knowledge Graph Embeddings

Automated Relational Meta-learning

PyTorch: An Imperative Style, High-Performance Deep Learning Library

Few-Shot Knowledge Graph Completion

Inductive Relation Prediction by Subgraph Reasoning

Relation Adversarial Network for Low Resource Knowledge Graph Completion

Composition-based Multi-Relational Graph Convolutional Networks

DRUM: End-To-End Differentiable Rule Mining On Knowledge Graphs

KG-BERT: BERT for Knowledge Graph Completion

KagNet: Knowledge-Aware Graph Networks for Commonsense Reasoning

Deep Graph Library: Towards Efficient and Scalable Deep Learning on Graphs

RoBERTa: A Robustly Optimized BERT Pretraining Approach

Iteratively Learning Embeddings and Rules for Knowledge Graph Reasoning

Interaction Embeddings for Prediction and Explanation in Knowledge Graphs

Logic Attention Based Neighborhood Aggregation for Inductive Knowledge Graph Embedding

Fine-Grained Evaluation of Rule- and Embedding-Based Systems for Knowledge Graph Completion

RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space

Improving Natural Language Inference Using External Knowledge in the Science Questions Domain

One-Shot Relational Learning for Knowledge Graphs

Representation Learning on Graphs with Jumping Knowledge Networks

Knowledge Graph Embedding: A Survey of Approaches and Applications

DeepPath: A Reinforcement Learning Method for Knowledge Graph Reasoning

A Simple Neural Attentive Meta-Learner

Convolutional 2D Knowledge Graph Embeddings

An End-to-End Model for Question Answering over Knowledge Base with Cross-Attention Combining Global Knowledge

Knowledge Transfer for Out-of-Knowledge-Base Entities : A Graph Neural Network Approach

A simple neural network module for relational reasoning

Modeling Relational Data with Graph Convolutional Networks

Prototypical Networks for Few-shot Learning

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

Differentiable Learning of Logical Rules for Knowledge Base Reasoning

ConceptNet 5.5: An Open Multilingual Graph of General Knowledge

Collaborative Knowledge Base Embedding for Recommender Systems

Complex Embeddings for Simple Link Prediction

Meta-Learning with Memory-Augmented Neural Networks

Matching Networks for One Shot Learning

Knowledge Graph Embedding via Dynamic Mapping Matrix

Learning Entity and Relation Embeddings for Knowledge Graph Completion

Embedding Entities and Relations for Learning and Inference in Knowledge Bases

Wikidata

Knowledge Graph Embedding by Translating on Hyperplanes

Translating Embeddings for Modeling Multi-relational Data

AMIE: association rule mining under incomplete evidence in ontological knowledge bases

Toward an Architecture for Never-Ending Language Learning

Freebase: a collaboratively created graph database for structuring human knowledge

INDIGO: GNN-Based Inductive Knowledge Graph Completion Using Pair-Wise Encoding

Meta Relational Learning for Few-Shot Link Prediction in Knowledge Graphs

CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge

MetaNetworks

Representing Text for Joint Embedding of Text and Knowledge Bases

Siamese Neural Networks for One-Shot Image Recognition

Q2 How does the performance of MorsE compare to baselines in conducting question answering for KGs in inductive settings?