Abstract
The increasing complexity of consumer preferences in the notebook market requires advanced methodologies to effectively analyze user sentiments and prioritize product features for strategic decision-making. Traditional market research methods often fail to capture real-time, spontaneous consumer feedback while lacking integration with expert knowledge of actual purchasing behavior. This study introduces a novel hybrid framework that systematically combines aspect-based sentiment analysis (ABSA) of social media data with expert evaluations to bridge the gap between expressed consumer preferences and actual purchase drivers. The methodology analyzes 329,091 Twitter posts from January 2023 to June 2024, covering seven major notebook brands. Using the PyABSA framework, consumer sentiments toward 16 key notebook attributes are extracted and analyzed. Expert evaluations, conducted through fuzzy logic defuzzification, assign importance weights based on observed purchasing patterns, with features subsequently ranked using the TOPSIS multi-criteria decision-making method. Findings reveal that price, display quality, CPU performance, RAM capacity, and design constitute the most influential factors in notebook purchasing decisions. Negative sentiments concentrate on cooling systems, battery chargers, and warranty services, indicating critical improvement areas. Brand-specific analysis demonstrates that display quality ranks highest for Lenovo, while price dominates concerns for Dell, HP, and Microsoft, validating distinct market positioning strategies. By integrating machine learning-based sentiment analysis with structured expert knowledge, this research provides manufacturers with a quantifiable, actionable framework for optimizing product development and marketing strategies. The methodology enables companies to prioritize genuinely purchase-influencing features while addressing critical pain points, enhancing competitive positioning in dynamic markets. Future research should expand to cross-cultural consumer behavior analysis and real-time sentiment tracking systems.