Journal of Theoretical and Applied Electronic Commerce Research

The rapid expansion of e-commerce has significantly influenced consumer purchasing behavior, making user reviews a critical source of product-related information. However, the large volume of low-quality and superficial reviews limits the ability to obtain reliable insights. This study aims to classify Turkish e-commerce reviews as either useful or useless, thereby highlighting high-quality content to support more informed consumer decisions. A dataset of 15,170 Turkish product reviews collected from major e-commerce platforms was analyzed using traditional machine learning approaches, including Support Vector Machines and Logistic Regression, and transformer-based models such as BERT and RoBERTa. In addition, a novel Multi-Transformer Fusion Framework (MTFF) was proposed by integrating BERT and RoBERTa representations through concatenation, weighted-sum, and attention-based fusion strategies. Experimental results demonstrated that the concatenation-based fusion model achieved the highest performance with an F1-score of 91.75%, outperforming all individual models. Among standalone models, Turkish BERT achieved the best performance (F1: 89.37%), while the BERT + Logistic Regression hybrid approach yielded an F1-score of 88.47%. The findings indicate that multi-transformer architectures substantially enhance classification performance, particularly for agglutinative languages such as Turkish. To improve the interpretability of the proposed framework, SHAP (SHapley Additive exPlanations) was employed to analyze feature contributions and provide transparent explanations for model predictions, revealing that the model primarily relies on experience-oriented and semantically meaningful linguistic cues. The proposed approach can support e-commerce platforms by automatically prioritizing high-quality and informative reviews, thereby improving user experience and decision-making processes.

This study examines why privacy concerns do not consistently deter online information disclosure by focusing on internal evaluative dynamics underlying privacy decisions. Drawing on theories of attitudinal ambivalence and cognitive–affective inconsistency, it investigates how internal tensions shape the translation of privacy concerns into disclosure behavior. Using two-phase data comprising a survey, the research distinguishes between threat-based and coping-based evaluative conflicts by operationalizing ambivalence and cognitive–affective inconsistency across privacy risks, perceived benefits, self-efficacy, and response efficacy. Results from Phase 1, based on 540 Amazon Mechanical Turk participants, indicate that while privacy concerns generally reduce disclosure intentions, this effect is significantly weakened when individuals experience higher levels of cognitive–affective inconsistency and ambivalence. Although ambivalence significantly reduces the magnitude of inconsistency, it has a limited influence on the moderating role of inconsistency. Phase 2 findings further show that under conditions of high ambivalence, cognitive–affective inconsistency related to self-efficacy exerts a significant effect in situation-specific disclosure contexts. By elucidating the dynamic interplay of the internal tensions, this study clarifies when and why privacy concerns fail to predict disclosure behavior and highlights the importance of incorporating internal evaluative dynamics into models of digital privacy decision-making.

This review examines existing research on virtual streamers in live streaming commerce and digital marketing, identifying key factors that shape consumer responses. Based on 41 peer-reviewed studies and following PRISMA 2020 guidelines, the analysis applies the CIMCO to synthesize findings through a systematic review. Results highlight three primary mechanisms—trait-based trust, perceived social presence, and message framing—which collectively constitute an integrative model explaining how virtual streamers influence AI-enabled consumer behavior. These elements shape how consumers engage with virtual streamers across platforms and product types. However, current research is limited by geographic concentration, reliance on self-reports, and a lack of longitudinal or behavioral data, which constrains broader applicability. For retailers and platform operators, aligning avatar traits and communication styles with product categories and consumer expectations is crucial for effective digital service delivery. Transparency about whether a streamer is AI or human-operated is also important for maintaining user trust. This review proposes a triadic integration model and offers a foundation for future research on AI-driven marketing influence.