Yoga Handoko Agustin, Nabil Nur Afrizal

SENTIMENT ANALYSIS OF IT WORKERS ON NO CODE AND LOW CODE TRENDS: COMPARISON OF LSTM AND SVM MODELS

Introduction

Sentiment analysis of it workers on no code and low code trends: comparison of lstm and svm models. Analyze IT professionals' sentiment on No Code & Low Code trends. Comparing LSTM & SVM, findings show cautious optimism and SVM's 87% accuracy.

Abstract

This research explores the sentiment of IT professionals toward the growing trend of No Code and Low Code technologies by comparing the performance of Support Vector Machine (SVM) and Long Short-Term Memory (LSTM) algorithms. Using the SEMMA methodology and automatic labeling with ChatGPT, a total of 4,238 comments were collected from Reddit and Twitter and categorized into positive, neutral, and negative sentiments. The analysis showed that neutral sentiment dominates on both platforms (47.9% on Reddit and 48.8% on Twitter), followed by positive sentiment (41.3% and 43.1%, respectively), indicating cautious but optimistic attitudes toward LCDPs. In terms of model performance, SVM outperformed LSTM with 87% accuracy and a weighted F1-score of 0.87, compared to LSTM’s 80% accuracy and a weighted F1-score of 0.80. These findings confirm that classical machine learning methods remain highly effective for short-text sentiment analysis in social media, particularly when combined with TF-IDF feature representation, SMOTE balancing, and LLM-based automatic labeling, while also offering new insights into IT community perceptions of disruptive technologies

Review

This research presents a timely and relevant exploration into the sentiment of IT professionals regarding No Code and Low Code (NCLC) technologies, a domain of increasing strategic importance. The paper effectively addresses a dual objective: quantifying IT worker sentiment and comparing the efficacy of classical machine learning (SVM) against deep learning (LSTM) for this specific short-text sentiment analysis task. Employing a robust methodology leveraging the SEMMA framework, social media data from Reddit and Twitter, and the innovative application of ChatGPT for automatic labeling, the study provides a comprehensive analytical approach to a critical industry trend. The findings offer valuable insights into both the IT community's perceptions and the practical application of sentiment analysis models. The observation of a dominant neutral sentiment, closely followed by positive attitudes, suggesting a "cautious but optimistic" stance, accurately reflects the nuanced and often debated impact of NCLC on IT roles. Furthermore, the head-to-head comparison of machine learning models yields a significant conclusion: Support Vector Machine, when augmented with TF-IDF feature representation, SMOTE balancing, and LLM-based labeling, demonstrably outperforms LSTM for short-text social media sentiment analysis. This result underscores the continued relevance and efficiency of classical methods in certain contexts, challenging the prevailing notion that deep learning always offers superior performance, particularly in data-constrained or short-text scenarios. While the study provides robust quantitative results, further qualitative exploration would significantly enrich the findings. Delving into the specific reasons underpinning the "cautious" aspect of the observed sentiment – perhaps through thematic analysis of negative comments or nuanced discussions of perceived challenges versus benefits – could offer more actionable insights for NCLC developers and implementers. Additionally, while SVM's superior performance is clear, a more elaborate discussion on the architectural implications and comparative computational costs of both models, particularly concerning their scalability with larger or more complex datasets, would add another layer of depth. Finally, given the innovative use of ChatGPT for automatic labeling, a brief discussion on its potential biases, limitations, or validation against human expert annotation would strengthen the methodological rigor and generalizability of the findings across diverse social media contexts.

Full Text

Comments

You need to be logged in to post a comment.