BriefGPT.xyz
Aug, 2018
从WordNet基于相似度度量学习图嵌入
Learning Graph Embeddings from WordNet-based Similarity Measures
HTML
PDF
Andrey Kutuzov, Alexander Panchenko, Sarah Kohail, Mohammad Dorgham, Oleksiy Oliynyk...
TL;DR
本研究提出path2vec，一种学习图嵌入的新方法，它依赖于节点相似性的结构度量，该模型使用一个密集的空间来学习节点的表示，以逼近用户定义的图距离度量，并在语义相似性和词义消歧任务方面表现出色。
Abstract
We present a new approach for learning
graph embeddings
, that relies on structural measures of
node similarities
for generation of training data. The model learns node embeddings that are able to approximate a gi
→