Aspect-based sentiment analysis of employee satisfaction and company recommendation in Malaysia’s it industry based on Glassdoor reviews

The Malaysian information technology (IT) sector has seen significant transformation, with job seekers increasingly relying on employee-generated content to assess company culture and workplace satisfaction. However, many existing platforms do not break down reviews into specific aspects, making it difficult for users to make employment decisions based on their preferences of aspects of employee satisfaction. This study addresses the need for structured insights by implementing aspect-based sentiment analysis (ABSA) on employee reviews sourced from Glassdoor. The system employs Support Vector Machine (SVM) classifiers to categorise reviews into sentiments, as well as into one of five aspects of employee satisfaction which are career growth, management, benefits, salary, and work-life balance. The classification models achieved 81.47% accuracy for sentiment classification and 75.64% for aspect classification. Moreover, a content-based filtering (CBF) approach, supported by cosine similarity, was used to generate personalised company recommendations based on user priority of each aspect of employee satisfaction. The system’s features include interactive data visualisations using charts and word clouds which provide users with an intuitive understanding of sentiment trends. Usability testing showed that the system reduced decision making time by 84.18%, with an average user satisfaction score of 4.93 out of 5. The findings demonstrate that applying ABSA and CBF to Glassdoor reviews can significantly improve career decision-making by offering structured, aspect-level insights and personalised recommendations.