Augmenting Compositional Models for Knowledge Base Completion Using Gradient Representations

2 Nov 2018 Matthias Lalisse Paul Smolensky

Neural models of Knowledge Base data have typically employed compositional representations of graph objects: entity and relation embeddings are systematically combined to evaluate the truth of a candidate Knowedge Base entry. Using a model inspired by Harmonic Grammar, we propose to tokenize triplet embeddings by subjecting them to a process of optimization with respect to learned well-formedness conditions on Knowledge Base triplets... (read more)

PDF Abstract
No code implementations yet. Submit your code now

Datasets


Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Link Prediction FB15k HHolE Hits@10 .901 # 1
Hits@1 .848 # 1
Hits@3 .727 # 2
MR 21 # 3
MRR .796 # 1
MRR filtered .796 # 1
Knowledge Graphs FB15k HHolE MRR .796 # 1
Link Prediction FB15k HHolE MR 21 # 2
MRR .796 # 8
Hits@10 .901 # 4
Hits@3 .848 # 2
Hits@1 .727 # 7
Link Prediction WN18 HHolE MRR .939 # 16
Hits@10 .951 # 15
Hits@3 .945 # 11
Hits@1 .931 # 14
MR 183 # 3

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet