Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Arabic Sentiment Analysis with Noisy Deep Explainable Model

Md. Atabuzzaman, Md Shajalal, Maksuda Bilkis Baby, Alexander Boden

TL;DR

This work tackles Arabic sentiment analysis by addressing data scarcity and the black-box nature of deep models. It introduces Gaussian noise layers to BiLSTM and CNN-BiLSTM architectures to reduce overfitting and employs LIME to provide local explanations for individual predictions. Evaluations on LABR and HTL show the noise-augmented models achieve competitive performance and improved interpretability, highlighting key sentiment cues through XAI. While not strictly state-of-the-art across all baselines, the approach offers robust generalization and practical explainability for Arabic SA in real-world deployments, with avenues for future enhancement via federated learning and broader XAI integration.

Abstract

Sentiment Analysis (SA) is an indispensable task for many real-world applications. Compared to limited resourced languages (i.e., Arabic, Bengali), most of the research on SA are conducted for high resourced languages (i.e., English, Chinese). Moreover, the reasons behind any prediction of the Arabic sentiment analysis methods exploiting advanced artificial intelligence (AI)-based approaches are like black-box - quite difficult to understand. This paper proposes an explainable sentiment classification framework for the Arabic language by introducing a noise layer on Bi-Directional Long Short-Term Memory (BiLSTM) and Convolutional Neural Networks (CNN)-BiLSTM models that overcome over-fitting problem. The proposed framework can explain specific predictions by training a local surrogate explainable model to understand why a particular sentiment (positive or negative) is being predicted. We carried out experiments on public benchmark Arabic SA datasets. The results concluded that adding noise layers improves the performance in sentiment analysis for the Arabic language by reducing overfitting and our method outperformed some known state-of-the-art methods. In addition, the introduced explainability with noise layer could make the model more transparent and accountable and hence help adopting AI-enabled system in practice.

Arabic Sentiment Analysis with Noisy Deep Explainable Model

TL;DR

This work tackles Arabic sentiment analysis by addressing data scarcity and the black-box nature of deep models. It introduces Gaussian noise layers to BiLSTM and CNN-BiLSTM architectures to reduce overfitting and employs LIME to provide local explanations for individual predictions. Evaluations on LABR and HTL show the noise-augmented models achieve competitive performance and improved interpretability, highlighting key sentiment cues through XAI. While not strictly state-of-the-art across all baselines, the approach offers robust generalization and practical explainability for Arabic SA in real-world deployments, with avenues for future enhancement via federated learning and broader XAI integration.

Abstract

Sentiment Analysis (SA) is an indispensable task for many real-world applications. Compared to limited resourced languages (i.e., Arabic, Bengali), most of the research on SA are conducted for high resourced languages (i.e., English, Chinese). Moreover, the reasons behind any prediction of the Arabic sentiment analysis methods exploiting advanced artificial intelligence (AI)-based approaches are like black-box - quite difficult to understand. This paper proposes an explainable sentiment classification framework for the Arabic language by introducing a noise layer on Bi-Directional Long Short-Term Memory (BiLSTM) and Convolutional Neural Networks (CNN)-BiLSTM models that overcome over-fitting problem. The proposed framework can explain specific predictions by training a local surrogate explainable model to understand why a particular sentiment (positive or negative) is being predicted. We carried out experiments on public benchmark Arabic SA datasets. The results concluded that adding noise layers improves the performance in sentiment analysis for the Arabic language by reducing overfitting and our method outperformed some known state-of-the-art methods. In addition, the introduced explainability with noise layer could make the model more transparent and accountable and hence help adopting AI-enabled system in practice.
Paper Structure (12 sections, 1 equation, 5 figures, 3 tables)

This paper contains 12 sections, 1 equation, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Proposed Method
  4. Classification Models
  5. Explainable Surrogate Model with LIME
  6. Experimental Results with Discussion
  7. Dataset
  8. Experimental setup
  9. Experimental results
  10. Comparison with related works
  11. XAI in Arabic Sentiment Analysis
  12. Conclusion and Future Work

Figures (5)

  • Figure 1: Overview diagram of our proposed Arabic Sentiment Analysis method
  • Figure 2: Explanation of a particular class (Positive) using LIME.
  • Figure 3: Explanation of a particular class (Negative) using LIME.
  • Figure 4: Explanation of a review with $Model_{ND}$ using LIME.
  • Figure 5: Explanation of a review with $Model_D$ using LIME.