Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Advancing Aspect-Based Sentiment Analysis through Deep Learning Models

Chen Li, Huidong Tang, Jinli Zhang, Xiujing Guo, Debo Cheng, Yasuhiko Morimoto

TL;DR

This work tackles ABSA by addressing long‑range and syntactic dependencies through an edge‑enhanced graph approach. SentiSys fuses Bi‑LSTM and Transformer with an edge‑enhanced Bi‑GCN guided by a dependency‑parsing layer and an aspect‑specific masking module, enabling precise aspect–sentiment alignment. Empirical results on four benchmark datasets show that SentiSys outperforms many baselines, including ASGCN, across multiple metrics, with peak gains observed when using three GCN layers. The approach demonstrates the value of incorporating syntactic edge information and targeted masking to improve aspect‑level sentiment classification, though it notes future work to optimize SDI computation and transformer integration for further gains.

Abstract

Aspect-based sentiment analysis predicts sentiment polarity with fine granularity. While graph convolutional networks (GCNs) are widely utilized for sentimental feature extraction, their naive application for syntactic feature extraction can compromise information preservation. This study introduces an innovative edge-enhanced GCN, named SentiSys, to navigate the syntactic graph while preserving intact feature information, leading to enhanced performance. Specifically,we first integrate a bidirectional long short-term memory (Bi-LSTM) network and a self-attention-based transformer. This combination facilitates effective text encoding, preventing the loss of information and predicting long dependency text. A bidirectional GCN (Bi-GCN) with message passing is then employed to encode relationships between entities. Additionally, unnecessary information is filtered out using an aspect-specific masking technique. To validate the effectiveness of our proposed model, we conduct extensive evaluation experiments on four benchmark datasets. The experimental results demonstrate enhanced performance in aspect-based sentiment analysis with the use of SentiSys.

Advancing Aspect-Based Sentiment Analysis through Deep Learning Models

TL;DR

This work tackles ABSA by addressing long‑range and syntactic dependencies through an edge‑enhanced graph approach. SentiSys fuses Bi‑LSTM and Transformer with an edge‑enhanced Bi‑GCN guided by a dependency‑parsing layer and an aspect‑specific masking module, enabling precise aspect–sentiment alignment. Empirical results on four benchmark datasets show that SentiSys outperforms many baselines, including ASGCN, across multiple metrics, with peak gains observed when using three GCN layers. The approach demonstrates the value of incorporating syntactic edge information and targeted masking to improve aspect‑level sentiment classification, though it notes future work to optimize SDI computation and transformer integration for further gains.

Abstract

Aspect-based sentiment analysis predicts sentiment polarity with fine granularity. While graph convolutional networks (GCNs) are widely utilized for sentimental feature extraction, their naive application for syntactic feature extraction can compromise information preservation. This study introduces an innovative edge-enhanced GCN, named SentiSys, to navigate the syntactic graph while preserving intact feature information, leading to enhanced performance. Specifically,we first integrate a bidirectional long short-term memory (Bi-LSTM) network and a self-attention-based transformer. This combination facilitates effective text encoding, preventing the loss of information and predicting long dependency text. A bidirectional GCN (Bi-GCN) with message passing is then employed to encode relationships between entities. Additionally, unnecessary information is filtered out using an aspect-specific masking technique. To validate the effectiveness of our proposed model, we conduct extensive evaluation experiments on four benchmark datasets. The experimental results demonstrate enhanced performance in aspect-based sentiment analysis with the use of SentiSys.
Paper Structure (20 sections, 11 equations, 2 figures, 2 tables)

This paper contains 20 sections, 11 equations, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Aspect-based Sentiment Analysis Systems
  4. RNN-based Sentiment Analysis Systems
  5. GCN-based Sentiment Analysis Systems
  6. SentiSys
  7. Word Embedding Layer
  8. Bi-LSTM Network
  9. Transformer Network
  10. Dependency-parsing Layer
  11. Bi-GCN Network
  12. Aspect-Specific Masking
  13. Sentiment Classification
  14. Experiments
  15. Datasets
  16. ...and 5 more sections

Figures (2)

  • Figure 1: SentiSys for aspect-based sentiment analysis includes the following components: (a) Initial word embedding layer for parsing and embedding words into vectors. (b) Bi-LSTM network captures contextual information for deeper word relationship understanding. (c) Transformer layer with self-attention mechanism explores global word features, enhancing contextual understanding. (d) Dependency parsing layer recognizes word dependencies, constructing a dependency tree to explain grammatical structure. (e) Bi-GCN propagates information among tree nodes, effectively modeling word relationships. (f) Aspect-specific masking refines hidden representation, reducing redundancy and improving accuracy. (g) Sentiment classification layer categorizes sentence sentiment into predefined categories, such as positive, negative, or neutral.
  • Figure 2: Change curve of Acc scores with the number of GCN layers.