• Publications
  • Influence
Composition-based Multi-Relational Graph Convolutional Networks
TLDR
This paper proposes CompGCN, a novel Graph Convolutional framework which jointly embeds both nodes and relations in a relational graph and leverages a variety of entity-relation composition operations from Knowledge Graph Embedding techniques and scales with the number of relations.
RESIDE: Improving Distantly-Supervised Neural Relation Extraction using Side Information
TLDR
RESIDE is a distantly-supervised neural relation extraction method which utilizes additional side information from KBs for improved relation extraction and employs Graph Convolution Networks to encode syntactic information from text and improves performance even when limited side information is available.
A Re-evaluation of Knowledge Graph Completion Methods
TLDR
This paper proposes a simple evaluation protocol that is robust to handle bias in the model, which can substantially affect the final results, and conducts extensive experiments and reports performance of several existing methods using the protocol.
CESI: Canonicalizing Open Knowledge Bases using Embeddings and Side Information
TLDR
Canonicalization using Embeddings and Side Information (CESI) is proposed -- a novel approach which performs canonicalization over learned embeddings of Open KBs by incorporating relevant NP and relation phrase side information in a principled manner.
Incorporating Syntactic and Semantic Information in Word Embeddings using Graph Convolutional Networks
TLDR
Word embeddings learned by SynGCN outperform existing methods on various intrinsic and extrinsic tasks and provide an advantage when used with ELMo and an effective framework for incorporating diverse semantic knowledge for further enhancing learned word representations are proposed.
InteractE: Improving Convolution-based Knowledge Graph Embeddings by Increasing Feature Interactions
TLDR
This paper analyzes how increasing the number of these interactions affects link prediction performance, and proposes InteractE, a feature permutation, a novel feature reshaping, and circular convolution approach that outperforms state-of-the-art convolutional link prediction baselines on FB15k-237.
Attention Interpretability Across NLP Tasks
TLDR
This work attempts to fill the gap by giving a comprehensive explanation which justifies both kinds of observations (i.e., when is attention interpretable and when it is not) and reinforces the claim of interpretability of attention through manual evaluation.
MedType: Improving Medical Entity Linking with Semantic Type Prediction
TLDR
This paper presents MedType, a fully modular system that prunes out irrelevant candidate concepts based on the predicted semantic type of an entity mention, and incorporates it into five off-the-shelf toolkits for medical entity linking and demonstrates that it consistently improves entity linking performance across several benchmark datasets.
Dating Documents using Graph Convolution Networks
TLDR
This paper proposes NeuralDater, a Graph Convolutional Network (GCN) based document dating approach which jointly exploits syntactic and temporal graph structures of document in a principled way and is the first application of deep learning for the problem of document dating.
Confidence-based Graph Convolutional Networks for Semi-Supervised Learning
TLDR
ConfGCN is proposed, which estimates labels scores along with their confidences jointly in GCN-based setting and uses these estimated confidences to determine the influence of one node on another during neighborhood aggregation, thereby acquiring anisotropic capabilities.
...
1
2
3
...