Performance Comparison of Static and Contextual Embedding Models for Opinion Mining
Open Access
- 27 April 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 65819-65832
Abstract
Transformer-based language models are widely adopted for sentiment analysis due to their strong contextual representation capabilities; however, their high computational complexity can limit practical deployment in resource-constrained environments. This work presents a systematic comparative study of static and contextual word representations for sentiment classification across architecturally distinct model families under standardized preprocessing and evaluation conditions. We evaluate Long Short-Term Memory (LSTM) networks with Word2Vec, GloVe, and FastText embeddings, and compare them with fine-tuned BERT-base and DistilBERT transformer baselines using the same preprocessing, training, and evaluation protocols across five publicly available datasets spanning formal reviews and noisy social media text. In addition to overall accuracy, we report macro- and weighted-F1 scores across three random seeds to estimate experimental variance, and apply McNemar’s test (with Bonferroni correction for multiple comparisons) to assess pairwise model differences on each dataset’s test set. Contingency tables required for each McNemar comparison are provided in the supplementary material. Detailed error analyses using confusion matrices and out-of-vocabulary (OOV) statistics examine the impact of lexical coverage on model performance. Experimental results indicate that FastText consistently outperforms other static embeddings, particularly on datasets with high lexical variability, while offering substantially lower computational overhead compared to the transformer baselines. The primary contribution of this study is a diagnostic multi-dataset evaluation framework that quantifies the relationship between lexical coverage and subword modeling benefit, and characterises the efficiency–accuracy trade-off across embedding paradigms. All architectural and training differences between model families are explicitly documented and their confounding effects discussed rather than assumed away. These findings provide practical guidance for selecting embedding strategies by jointly considering accuracy and deployment efficiency.Keywords
This publication has 26 references indexed in Scilit:
- Harnessing CNNs and Embedding Techniques for Enhanced Sentiment ClassificationNational Academy Science Letters, 2025
- Strategies for enhancing the performance of news article classification in Bangla: Handling imbalance and interpretationEngineering Applications of Artificial Intelligence, 2023
- Social media-based opinion retrieval for product analysis using multi-task deep neural networksExpert Systems with Applications, 2021
- Sentiment analysis of social media data based on chaotic coyote optimization algorithm based time weight‐AdaBoost support vector machine approachConcurrency and Computation: Practice and Experience, 2021
- Lexicon-based sentiment analysis using Twitter data2021
- Sentiment Analysis Using Gated Recurrent Neural NetworksSN Computer Science, 2020
- Sentiment analysis on IMDB using lexicon and neural networksSN Applied Sciences, 2020
- Sentiment analysis algorithms and applications: A surveyAin Shams Engineering Journal, 2014
- An empirical study of sentiment analysis for chinese documentsExpert Systems with Applications, 2008
- Note on the Sampling Error of the Difference Between Correlated Proportions or PercentagesPsychometrika, 1947