<![CDATA[Purpose of the study: This study investigates the transformative role of Meaning Reconstruction (MR) in mediating the relationship between Post-Purchase Dissonance (PPD) and Eudaimonic Well-Being (EWB) in the context of digital education service consumption in Indonesia. Methodology: This study employs a quantitative explanatory design using Structural Equation Modeling–Partial Least Squares (SEM-PLS). Data were collected through web scraping from 500 publicly accessible posts on Twitter (X), Reddit, and TikTok discussing post-purchase experiences with digital education services in Indonesia (2020–2024). Computational text analysis using a fine-tuned IndoBERT model was applied to derive construct scores for PPD (X1–X3), MR (M1–M4), and EWB (Y1–Y4). Indicators were quantified using algorithmically normalized NLP-based semantic intensity scores ranging from 1 to 7, representing computational estimations of construct magnitude rather than self-reported NLP-based normalized semantic score (1–7) responses. Main Findings: Results confirm that PPD significantly influences MR (β=0.446, p<0.001) and MR strongly predicts EWB (β=0.481, p<0.001). The direct path PPD→EWB is significant (β=0.176, p<0.001), while the indirect effect via MR is also significant (β=0.214, t=8.248, p<0.001), with a Variance Accounted For (VAF) of 54.8%, indicating partial mediation. R²(MR)=0.199 and R²(EWB)=0.338. All measurement model criteria are satisfied: outer loadings >0.70, AVE >0.50, CR >0.70, and HTMT <0.85. Novelty/Originality of this study: Reconstruction as a critical transformative mechanism, reframing post-purchase dissonance from a purely negative outcome into a catalyst for consumer psychological growth. The use of computational text analysis (NLP + web scraping) as an alternative to conventional surveys further contributes methodological novelty.]]>
