Traversing Knowledge Graphs in Vector Space

Abstract

Path queries on a knowledge graph can be used to answer compositional questions such as "What languages are spoken by people living in Lisbon?". However, knowledge graphs often have missing facts (edges) which disrupts path queries. Recent models for knowledge base completion impute missing facts by embedding knowledge graphs in vector spaces. We show that these models can be recursively applied to answer path queries, but that they suffer from cascading errors. This motivates a new "compositional" training objective, which dramatically improves all models' ability to answer path queries, in some cases more than doubling accuracy. On a standard knowledge base completion task, we also demonstrate that compositional training acts as a novel form of structural regularization, reliably improving performance across all base models (reducing errors by up to 43%) and achieving new state-of-the-art results.

2015 Conference on Empirical Methods on Natural Language Processing (EMNLP)

References in corpus (3)

• Theano: new features and speed improvements
• Question Answering with Subgraph Embeddings
• Compositional Vector Space Models for Knowledge Base Completion

Cited by in corpus (52)

• Comparative Study of CNN and RNN for Natural Language Processing
• A Novel Embedding Model for Knowledge Base Completion Based on Convolutional Neural Network
• Analogical Inference for Multi-Relational Embeddings
• DeepPath: A Reinforcement Learning Method for Knowledge Graph Reasoning
• An Interpretable Reasoning Network for Multi-Relation Question Answering
• Modeling Relational Data with Graph Convolutional Networks
• Representation Learning for Natural Language Processing
• Knowledge Transfer for Out-of-Knowledge-Base Entities: A Graph Neural Network Approach
• Analysis of the Impact of Negative Sampling on Link Prediction in Knowledge Graphs
• KBLRN : End-to-End Learning of Knowledge Base Representations with Latent, Relational, and Numerical Features
• Neighborhood Mixture Model for Knowledge Base Completion
• Natural Language Processing for Information Extraction
• Multi-View Graph Neural Networks for Molecular Property Prediction
• Improving Information Extraction by Acquiring External Evidence with Reinforcement Learning
• Thinking, Fast and Slow: Combining Vector Spaces and Knowledge Graphs
• On the Equivalence of Holographic and Complex Embeddings for Link Prediction
• Learning multi-faceted representations of individuals from heterogeneous evidence using neural networks
• CFO: Conditional Focused Neural Question Answering with Large-scale Knowledge Bases
• Data Recombination for Neural Semantic Parsing
• The More You Know: Using Knowledge Graphs for Image Classification
• Knowledge Graph Embedding with Iterative Guidance from Soft Rules
• Knowledge Representation Learning: A Quantitative Review
• Faithful Embeddings for Knowledge Base Queries
• Simpler Context-Dependent Logical Forms via Model Projections
• Gaussian Attention Model and Its Application to Knowledge Base Embedding and Question Answering
• Cross-Domain Aspect Extraction using Transformers Augmented with Knowledge Graphs
• Learning Sequence Encoders for Temporal Knowledge Graph Completion
• Weakly-supervised Contextualization of Knowledge Graph Facts
• Knowledge Graph Embedding with Entity Neighbors and Deep Memory Network
• Incorporating Relation Paths in Neural Relation Extraction
• Inferring Logical Forms From Denotations
• A Neural Language Model for Dynamically Representing the Meanings of Unknown Words and Entities in a Discourse
• Learn to Explain Efficiently via Neural Logic Inductive Learning
• Predicting ConceptNet Path Quality Using Crowdsourced Assessments of Naturalness
• Rule-Guided Compositional Representation Learning on Knowledge Graphs
• Link Prediction using Embedded Knowledge Graphs
• On Evaluating Embedding Models for Knowledge Base Completion
• Dynamic Anticipation and Completion for Multi-Hop Reasoning over Sparse Knowledge Graph
• Compositional Learning of Relation Path Embedding for Knowledge Base Completion
• A Survey of Embedding Space Alignment Methods for Language and Knowledge Graphs
• Seeing is Knowing! Fact-based Visual Question Answering using Knowledge Graph Embeddings
• An Adversarial Transfer Network for Knowledge Representation Learning
• Learning Semantically and Additively Compositional Distributional Representations
• Recurrent One-Hop Predictions for Reasoning over Knowledge Graphs
• Information Extraction from Scientific Literature for Method Recommendation
• Semantically Enhanced Models for Commonsense Knowledge Acquisition
• The Profiling Machine: Active Generalization over Knowledge
• Improved Knowledge Base Completion by Path-Augmented TransR Model
• Recent Advances in Automated Question Answering In Biomedical Domain
• R4C: A Benchmark for Evaluating RC Systems to Get the Right Answer for the Right Reason
• Dynamic Adaptive Network Intelligence
• Navigation-Based Candidate Expansion and Pretrained Language Models for Citation Recommendation