<![CDATA[This study addresses the need to understand the dynamics of sentiment and social network analysis (SNA) in the context of smartwatch product reviews. Leveraging the CRoss-Industry Standard Process for Data Mining (CRISP-DM) methodology, the research aims to analyze sentiments and social networks to glean insights into consumer behavior and interaction patterns. The CRISP-DM framework guides the research through structured phases of business understanding, data understanding, data preparation, modeling, evaluation, and deployment. Through sentiment analysis using Support Vector Machine (SVM) with Synthetic Minority Over-sampling Technique (SMOTE) and SNA, the study examines accuracy (91.41% +/- 1.66%), precision (100.00% +/- 0.00%), recall (82.80% +/- 3.36%), f-measure (90.56% +/- 2.01%), Area Under the Curve (AUC), as well as network metrics such as diameter (4), density (0.001036), reciprocity (0.000000), centralization (0.004920), and modularity (0.994200). Findings reveal a robust performance of the SVM algorithm coupled with SMOTE, showcasing high accuracy and effective discrimination between sentiments. Additionally, SNA uncovers valuable insights into network structures, communication patterns, and sentiment propagation dynamics within the online community. These findings contribute to a deeper understanding of consumer sentiments and interactions, guiding strategic marketing, product development, and reputation management decisions.]]>
Copyrights © 2024
