.png){kind=icon}
Return
A Multiple Graphical Convolution Networks Approach for Aspect-Based Sentiment Classification
.png){kind=icon}
PRE
AI
DOI：10.1007/978-3-031-97141-9_5.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Aspect-Based Sentiment Classification (ABSC) has attracted a lot of attention in recent years. Some of the best performing models for ABSC are the hybrid ones and the ones based on deep learning. This work presents a model for ABSC dubbed HAABSA4GCN, as an extension to the state-of-the-art hybrid model HAABSA++. The HAABSA4GCN model combines a domain ontology from HAABSA++ with a novel backup model based on the Graph Convolution Neural (GCN) network, dubbed 4GCN. This backup model uses four graphs: syntactic, semantic, lexical, and ontological. For the evaluation, the SemEval 2015 and SemEval 2016 datasets are used. The evaluation results show that HAABSA4GCN outperforms HAABSA++, suggesting that the combination of different graphs within the model increases the prediction accuracy. In addition, 4GCN performs better than HAABSA4GCN, implying that 4GCN is better in dealing with cases that the domain ontology can decide upon.
Keywords:
Aspect-based sentiment classification
Graph convolution neural network
Domain ontology
Journal
N
IF:
0
Papers:
30
Citations:
0
Organization
Cited Papers
.png){kind=icon}
ShareSaveShareSave
View more