AI-based Sentiment Analysis for Enhancing Employee Satisfaction using GRU-Attention Mechanism

Employee satisfaction is crucial for organizational success, and understanding employee sentiment through feedback analysis can provide actionable insights. This research proposes an AI-based sentiment analysis model using the GRU-Attention mechanism to enhance employee satisfaction. Data collected from Kaggle includes employee feedback from various organizations, which undergoes preprocessing using techniques such as tokenization, stopword removal, and lemmatization. Text features are extracted using the TF-IDF method, which converts textual data into numeric form. The GRU-Attention model leverages the strengths of gated recurrent units (GRUs) and attention mechanisms to focus on key phrases in employee feedback for sentiment classification into positive, negative, or neutral categories. Results demonstrate that the proposed model achieves higher accuracy (91.8%) compared to baseline methods like Random Forest (82.1%) and GRU without Attention (88.5%). Additionally, the GRU-Attention model outperforms a traditional LSTM-Attention model, showcasing its efficiency in sentiment classification. The analysis also highlights the importance of focusing on specific feedback elements that influence employee sentiment. The findings suggest that integrating AI-based sentiment analysis into organizational workflows can improve employee engagement, workplace satisfaction, and decision-making. The model's performance in terms of training time and computational efficiency further demonstrates its practical viability for real-world applications. Future research could explore additional sentiment nuances and apply the model across different domains for further generalization.