search

Cartus / AGGCN

Attention Guided Graph Convolutional Networks for Relation Extraction (authors' PyTorch implementation for the ACL19 paper)
MIT License
432 stars 88 forks source link

Dependency type #15

Closed wt123u closed 4 years ago

wt123u commented 4 years ago

Hi @Cartus Thank you very much for your sharing. I would like to ask whether the dependency type information is used or whether only the edge between the nodes is considered and the specific edge type is not considered?

Cartus commented 4 years ago

Hi @wt123u,

Yes. The edge type is not considered in this model. In this work, we only make use of the connection information (i.e. which two tokens are connected).

Other GNN models considered the edge types (N-ary Relation Extraction using Graph State LSTM, Song, et al 2018). What they did is basically concatenate the edge embedding to the node embedding during the information propagation process. You can check their code and paper.

wt123u commented 4 years ago

Hi @Cartus Thanks so much for your prompt reply. I will see Graph State LSTM soon.