Decoding Sentiments: Harnessing the Power of NLP for Comparative Analysis of ML Algorithms

Machine learning algorithms have become pervasive in diverse applications, revolutionizing various domains. However, the abundance of algorithms, each designed for specific purposes, poses a challenge for both novice users and experts in selecting the most suitable model. This research addresses this issue through a comprehensive analysis, leveraging Natural Language Processing (NLP) techniques and the powerful Sentiment decoding libraries. Researcher’s comments from published papers were analyzed using the various approaches, presenting a novel approach that maps algorithm scores based on adjectives. Results indicate that Support Vector Machine consistently outperforms other algorithms, providing valuable insights for practitioners. Our research contributes to a systematic evaluation framework, aiding researchers in algorithm selection. This paper presents a study focused on sentiment analysis, aimed at identifying the most effective machine learning algorithm among various contenders. Traditional conclusions regarding the superiority of specific algorithms are often based on their performance on specific datasets or tasks, which may not provide a comprehensive comparison due to the diverse nature of algorithms. To address this, a standard dataset encompassing the advantages and disadvantages of all algorithms is utilized. Three approaches employing VADER, TextBlob, and transformers are employed for sentiment analysis. Results consistently highlight SVM as the top-performing algorithm, followed by Associative analysis. This standardized conclusion contributes to the broader understanding of sentiment analysis within the realm of machine learning algorithms.