Purpose/Significance Consumer psychology and demand preferences embedded within user reviews constitute core intelligence resources for precise business operations. This study focuses on the online book consumption scenario, aiming to construct an analytical framework for online book user purchase behavior based on Event Logic Graphs (ELGs). This framework deeply analyzes the internal logical chains and pattern regularities within user behavior events. It seeks to expand the research boundaries of user behavior analysis and ELG applications theoretically, while simultaneously providing practical support for e-commerce platform intelligent operations and the publishing industry’s precision marketing. Thus, it possesses both theoretical innovation value and application prospects. Methods/Process Using Dangdang.com book reviews as the data source, the Top2Vec unsupervised topic clustering method was employed to extract user purchase behavior themes. Combining this with Gephi, an ELG was constructed where clustered themes served as nodes and semantic relationships between themes as edges. Visualization techniques were leveraged to deduce the logic behind user purchase behavior and uncover latent demand preferences. Results/Conclusions Online book purchasing behavior exhibits a causal logical chain of “Motivation Triggering → Decision Implementation → Feedback Iteration”: The motivation layer encompasses diverse demand orientations like cognitive enhancement and emotional connection; the decision layer is driven by multidimensional factors including product aesthetics, social trust, and price perception; the feedback layer forms a closed-loop mechanism involving quality supervision and emotional continuity. Based on the behavioral characteristics revealed by the ELG, online book retailers need to anchor demand scenarios, building a precision operation system across four dimensions—content ecosystem, product form, marketing reach, and service quality control—to synergistically achieve growth in both user value and commercial value. Innovation/Limitation The innovation lies in integrating Top2Vec theme clustering with ELG visualization technology, establishing a “semantic aggregation + logical deduction” research paradigm for consumer behavior. Limitations include the potential for small sample themes to weaken the explanatory power for group heterogeneity, and the research scope being currently confined to the “purchase behavior” stage without extending to the entire reading lifecycle. Future research should deepen conclusions by expanding data dimensions and scenario boundaries.
zhang et al. (2026) studied this question.