Multi-refined Feature Enhanced Sentiment Analysis Using Contextual Instruction
2025/11/01 by Atandoh, Peter, Zou, Jie, Guo, Weikang +2
#Computation and Language (cs.CL) #FOS: Computer and information sciences #Machine Learning (cs.LG)
paper · doi:10.48550/arxiv.2511.00537
Abstract
Sentiment analysis using deep learning and pre-trained language models (PLMs) has gained significant traction due to their ability to capture rich contextual representations. However, existing approaches often underperform in scenarios involving nuanced emotional cues, domain shifts, and imbalanced sentiment distributions. We argue that these limitations stem from inadequate semantic grounding, poor generalization to diverse linguistic patterns, and biases toward dominant sentiment classes. To overcome these challenges, we propose CISEA-MRFE, a novel PLM-based framework integrating Contextual Instruction (CI), Semantic Enhancement Augmentation (SEA), and Multi-Refined Feature Extraction (MRFE). CI injects domain-aware directives to guide sentiment disambiguation; SEA improves robustness through sentiment-consistent paraphrastic augmentation; and MRFE combines a Scale-Adaptive Depthwise Encoder (SADE) for multi-scale feature specialization with an Emotion Evaluator Context Encoder (EECE) for affect-aware sequence modeling. Experimental results on four benchmark datasets demonstrate that CISEA-MRFE consistently outperforms strong baselines, achieving relative improvements in accuracy of up to 4.6% on IMDb, 6.5% on Yelp, 30.3% on Twitter, and 4.1% on Amazon. These results validate the effectiveness and generalization ability of our approach for sentiment classification across varied domains.
Citations
- An Algorithm for Routing Vectors in Sequences
- Adaptive Prompt Learning-based Few-Shot Sentiment Analysis
- PromDA: Prompt-based Data Augmentation for Low-Resource NLU Tasks
- Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing
- Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing
- CoAtNet: Marrying Convolution and Attention for All Data Sizes
- The Power of Scale for Parameter-Efficient Prompt Tuning
- Discovering Emotion and Reasoning its Flip in Multi-Party Conversations using Masked Memory Network and Transformer
- GPT Understands, Too
- Making Pre-trained Language Models Better Few-shot Learners
- Exploiting Cloze Questions for Few Shot Text Classification and Natural\n Language Inference
- BART: Denoising Sequence-to-Sequence Pre-training for Natural Language\n Generation, Translation, and Comprehension
- DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
- Language Models as Knowledge Bases?
- RoBERTa: A Robustly Optimized BERT Pretraining Approach
- XLNet: Generalized Autoregressive Pretraining for Language Understanding
- How to Fine-Tune BERT for Text Classification?
- Utilizing BERT for Aspect-Based Sentiment Analysis via Constructing Auxiliary Sentence
- Pay Less Attention with Lightweight and Dynamic Convolutions
- Emotion Representation Mapping for Automatic Lexicon Construction (Mostly) Performs on Human Level
- MemNet: A Persistent Memory Network for Image Restoration
- Depthwise Separable Convolutions for Neural Machine Translation
- MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
- Comparative Study of CNN and RNN for Natural Language Processing
- Deep Residual Learning for Image Recognition
- Effective LSTMs for Target-Dependent Sentiment Classification
- A C-LSTM Neural Network for Text Classification
- Character-level Convolutional Networks for Text Classification
- Convolutional Neural Networks for Sentence Classification
Related