A Comparison of SVM against Pre-trained Language Models (PLMs) for Text Classification Tasks
This work addresses practitioners in NLP by showing that traditional methods like SVM may still be effective for text classification, but it is incremental as it compares existing methods without introducing new techniques.
The paper compared pre-trained language models (PLMs) against SVM with TFIDF for text classification, finding that PLMs, even when fine-tuned, did not significantly outperform SVM on domain-specific and public datasets, suggesting SVM can be cheaper and superior.
The emergence of pre-trained language models (PLMs) has shown great success in many Natural Language Processing (NLP) tasks including text classification. Due to the minimal to no feature engineering required when using these models, PLMs are becoming the de facto choice for any NLP task. However, for domain-specific corpora (e.g., financial, legal, and industrial), fine-tuning a pre-trained model for a specific task has shown to provide a performance improvement. In this paper, we compare the performance of four different PLMs on three public domain-free datasets and a real-world dataset containing domain-specific words, against a simple SVM linear classifier with TFIDF vectorized text. The experimental results on the four datasets show that using PLMs, even fine-tuned, do not provide significant gain over the linear SVM classifier. Hence, we recommend that for text classification tasks, traditional SVM along with careful feature engineering can pro-vide a cheaper and superior performance than PLMs.