Ontology Based Emotional Classification and Influence Analysis of Online Travel Agency User Opinions

Abstract

Rapid growth of social media discussions around online travel services generates large volumes of emotionally charged user opinions with measurable social impact. Interest arises from the need to interpret emotional signals while identifying influential actors shaping public perception. Proposed research addresses limitations of polarity focused opinion mining by unifying emotion recognition, service aspect semantics, and user influence assessment within Chines and Malaysian language social media data. The scope covers major online travel platforms and concentrates on financial services, booking support, platform experience, and event related interactions. The approach combines domain specific semantic modeling, deep contextual language representation, and graph based influence quantification supported by multi criteria decision analysis. Evidence includes manually validated emotion annotations, large scale interaction networks, and comparative performance evaluation across classification tasks. Findings show strong reliability for sentiment and service aspect detection with moderate performance for fine grained emotion categories, while influence ranking highlights structural network position as the primary determinant of opinion spread with engagement acting as an amplification signal. The integration of emotional semantics with influence scoring enriches interpretation of service related dissatisfaction and satisfaction patterns. Contributions extend opinion mining research for low resource languages and support practical applications in service monitoring, influencer identification, and targeted response strategies. Future work includes expanding emotion taxonomies, cross platform analysis, and longitudinal influence modeling.