Research And Application Of Aspect-Based Sentiment Analysis Using Graph Convolutional Network

The user-generated text content on the mobile Internet platforms contains potentially important commercial value and decision-making basis.Sentiment analysis and opinion mining of the content is a topic worth exploring.Different from traditional document-level and sentence-level sentiment analysis,Aspect-Based Sentiment Analysis(ABSA)performs more fine-grained sentiment and opinion mining based on entities and their aspects.This dissertation focuses on two sub-tasks of ABSA,Aspect Sentiment Classification(ASC)and Aspect Sentiment Triplet Extraction(ASTE).The main research work in this dissertation is as follows:1)In the ASC task,graph convolutional neural networks over dependency trees have been explored to explicitly model connections between aspects and opinion words.However,the improvement is limited due to the inaccuracy of the dependency parsing results and the informal expressions and complexity of online reviews.To overcome these challenges,this dissertation proposes a Dual Graph Convolutional Networks(DualGCN)model that considers the complementarity of syntactic structures and semantic correlations simultaneously.Particularly,to alleviate dependency parsing errors,this dissertation designs a SynGCN module with rich syntactic knowledge.To capture semantic correlations,this dissertation designs a SemGCN module with self-attention mechanism.Furthermore,this dissertation proposes orthogonal and differential regularizers to capture semantic correlations between words precisely by constraining attention scores in the SemGCN module.The orthogonal regularizer encourages the SemGCN to learn semantically correlated words with less overlap for each word.The differential regularizer encourages the SemGCN to learn semantic features that the SynGCN fails to capture.Experimental results on three public datasets show that DualGCN model outperforms state-of-the-art methods and verify the effectiveness of the model.2)In the ASTE task,most of the existing studies focus on devising a new tagging scheme that enables the model to extract the sentiment triplets in an end-to-end fashion.However,these methods ignore the relations between words for ASTE task.This dissertation proposes an Enhanced Multi-Channel Graph Convolutional Network model(EMC-GCN)to fully utilize the relations between words.Specifically,this dissertation first defines ten types of relations for ASTE task,and then adopt a biaffine attention module to embed these relations as an adjacent tensor between words in a sentence.After that,EMC-GCN transforms the sentence into a multi-channel graph by treating words and the relation adjacent tensor as nodes and edges,respectively.Thus,relation-aware node representations can be learnt.Furthermore,this dissertation considers diverse linguistic features to enhance EMC-GCN model.Finally,this dissertation designs an effective refining strategy on EMC-GCN for word-pair representation refinement,which considers the implicit results of aspect and opinion extraction when determining whether word pairs match or not.Extensive experimental results on the benchmark datasets demonstrate that the effectiveness and robustness of the proposed model,which outperforms cutting-edge methods significantly.3)Based on the study of ASC and ASTE tasks,this dissertation designs and implements an ABSA system.The system supports the selection of models in many different domains,online operations,and file data upload,which is simple to use and robust in performance.