E2EG: End-to-End Node Classification Using Graph Topology and Text-based Node Attributes

08/09/2022
by   Tu Anh Dinh, et al.
0

Node classification utilizing text-based node attributes has many real-world applications, ranging from prediction of paper topics in academic citation graphs to classification of user characteristics in social media networks. State-of-the-art node classification frameworks, such as GIANT, use a two-stage pipeline: first embedding the text attributes of graph nodes then feeding the resulting embeddings into a node classification model. In this paper, we eliminate these two stages and instead develop an end-to-end node classification model that builds upon GIANT, called End-to-End-GIANT (E2EG). The tandem utilization of a main and an auxiliary classification objectives in our approach results in a more robust model, thus enabling the BERT backbone to be switched out for a distilled encoder with a 25 number of parameters. Moreover, the end-to-end nature of the model increases ease of use, as it avoids the need of chaining multiple models for node classification. Compared to a GIANT+MLP baseline on the ogbn-arxiv and ogbn-products datasets, our model is able to obtain slightly better accuracy in the transductive setting (+0.5 40 GIANT+MLP by up to +2.23

READ FULL TEXT

page 1

page 2

page 3

page 4

research
10/22/2020

Joint Use of Node Attributes and Proximity for Semi-Supervised Classification on Graphs

The node classification problem is to infer unknown node labels in a gra...
research
02/11/2019

Deep Node Ranking: an Algorithm for Structural Network Embedding and End-to-End Classification

Complex networks are used as an abstraction for systems modeling in phys...
research
08/14/2017

Graph Classification via Deep Learning with Virtual Nodes

Learning representation for graph classification turns a variable-size g...
research
12/27/2019

Predicting Attributes of Nodes Using Network Structure

In many graphs such as social networks, nodes have associated attributes...
research
03/22/2020

Deep Multi-attributed Graph Translation with Node-Edge Co-evolution

Generalized from image and language translation, graph translation aims ...
research
03/17/2020

Author2Vec: A Framework for Generating User Embedding

Online forums and social media platforms provide noisy but valuable data...
research
05/24/2020

Integrated Node Encoder for Labelled Textual Networks

Voluminous works have been implemented to exploit content-enhanced netwo...

Please sign up or login with your details

Forgot password? Click here to reset