HPHeejun Park

Heejun Park

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Heejun Park has a total of 45 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Training-Label Efficiency in Vision-Based Inspection of LNG Tank Cables: A Duplicate-Aware and Calibration-Controlled Benchmark of Supervised and Normal-Only Anomaly Detection

    by Youngju Cho, Jiyoon Hong, Heejun Park

    Vision-based sensing is increasingly important for liquefied natural gas (LNG) tank manufacturing, where missed defects in cable-related components can affect monitoring, control, and safety functions. However, industrial image datasets often contain duplicate or near-duplicate imagery, operating thresholds can be optimistically tuned on held-out test data, and abnormal labels are costly to obtain. This study evaluates LNG tank cable defect detection under a duplicate-aware and calibration-controlled protocol that separates model training, threshold calibration, and held-out testing. Using the AI-Hub LNG tank quality inspection image dataset, we compare ResNet-18, EfficientNet-B0, and ConvNeXt-Tiny supervised classifiers; PatchCore, PaDiM, and STFPM normal-only training anomaly detectors; and ResNet-18 training-label-budget experiments that retain all normal training images while varying the number of labeled abnormal training images used for model fitting. Across 36 de-duplicated final runs, EfficientNet-B0 achieved the strongest fully supervised performance (AUROC 0.9999±0.0001, F1 0.9956±0.0029). PatchCore was the best normal-only training anomaly detector (AUROC 0.8909±0.0109, F1 0.8666±0.0228). In the percentage-budget experiment, 1% labeled abnormal training samples already exceeded PatchCore in AUROC, F1, and false-negative rate under the fixed labeled-calibration protocol; in a finer 4–64 abnormal-training-label sweep with 90 runs, the first training budget exceeding PatchCore across AUROC, F1, and false-negative rate with 16 labeled abnormal training images. These results quantify abnormal training-label efficiency under a fixed labeled-calibration protocol, rather than the total abnormal annotation burden required to deploy the complete operating classifier.

  • Open Access

    Badge Tenure as a Moderator of Review Cues: An Elaboration Likelihood Model Perspective on Yelp’s Elite Reviewers

    by Youngju Cho, Junyoung Yoo, Joon-Woo Yoo, Heejun Park

    Online reviews are increasingly pivotal in consumer decision-making, with platforms employing mechanisms such as badges to denote reviewer credibility. Although prior research has examined the influence of expert reviewers, it has typically treated badge holders as a homogeneous group, overlooking how variation in tenure within expert tiers shapes the way readers process review content. This article examines how Yelp Elite badge tenure, operationalized as Red (1–4 years), Gold (5–9 years), and Black (10+ years) tiers and treated as a proxy for accumulated platform-recognized expertise, moderates the effects of peripheral cues (Extremity, Length) and central cues (Readability, Subjectivity, and Plutchik’s eight emotions) on perceived helpfulness within an Elaboration Likelihood Model (ELM) framework. The analysis draws on the full population of 324,426 restaurant reviews authored by Yelp Elite badge holders between 2019 and 2021, using a pooled count-model specification with badge tier as a categorical moderator. The primary specification is estimated using Poisson quasi-maximum likelihood with HC1-robust standard errors, and full negative binomial estimation is reported as a robustness check. Wald tests indicate that badge tenure significantly moderates eight of twelve cue–helpfulness relationships (χ2(24)=4938, p<0.001). The effect of readability is monotonically positive and increases sharply with tenure, while the effect of joy varies across tenure groups. These findings suggest that reviewer expertise signals are not monolithic, refining theoretical insights on how tenure-based credibility cues moderate cue processing and offering practical implications for review platform management. The findings also indicate that platforms applying uniform ranking or surfacing rules across all Elite reviewers risk misallocating visibility, and that tenure-conditional weighting of textual cues warrants consideration.

  • Open Access

    A Study on the Impact of Corporate Mecenat Activities on Relationship Quality and Corporate Legitimacy

    by Jiah Hwang, Heejun Park

    Purpose: This study aims to empirically examine how different types of corporate mecenat activities (value oriented, stakeholder oriented, performance oriented) influence relationship quality and corporate legitimacy. It also seeks to identify the mediating roles of relationship quality and legitimacy within these relationships.Methods: The study categorizes corporate mecenat activities into three types and statistically tests their direct and mediating effects on relationship quality and corporate legitimacy using empirical analysis based on survey data.Results: Value oriented and stakeholder oriented mecenat activities were found to have significant positive effects on both relationship quality and corporate legitimacy. Performance oriented activities positively affected corporate legitimacy but did not show a significant effect on relationship quality. Similar patterns appeared in the mediating pathways among CMA types, relationship quality, and corporate legitimacy.Conclusion: Corporate mecenat activities function as adaptive strategies for firms responding to evolving societal expectations and can contribute to relationship stability and corporate legitimacy. However, due to sample characteristics, the findings may not generalize to all types of CMA. The study emphasizes the need for broader models applicable across diverse organizational and cultural contexts and offers implications for stakeholder relationship building, CRM, and CEM strategies.


  • Open Access

    Understanding the Impact of Inconsistency on the Helpfulness of Online Reviews

    by Junsung Park, Heejun Park

    This study investigates how review inconsistency influences perceived helpfulness in online restaurant reviews both in ratings and specific aspects of service attributes. Drawing on 106,464 Yelp reviews spanning 666 restaurants, we employed aspect-based sentiment analysis and Tobit regression to capture not only rating inconsistencies but also differences in sentiment toward décor, taste, service, and price. Results indicate that rating inconsistency negatively affects review helpfulness, suggesting that highly divergent ratings reduce credibility. However, aspect inconsistency shows mixed effects. Discrepancies in décor and taste positively influence perceived helpfulness by offering novel information, whereas service-related inconsistencies diminish review helpfulness, due to heightened consumer sensitivity to possible service failures. Reviewer expertise further strengthens the negative influence of inconsistency as readers expect experienced reviewers to provide objective feedback. These findings extend current research by shifting the analytical lens from individual reviews to sets of reviews, thereby capturing the relational dynamics that shape consumers’ perceptions of review credibility. The results also highlight the importance of analyzing review content by specific aspects to uncover nuanced effects. Practically, platforms can benefit from grouping reviews by attributes and alerting users to noteworthy inconsistencies, facilitating more informed consumer decision-making.


  • Open Access

    Analysis of Structural Changes in the Mobile-base E-commerce Market

    by Mi Yeon Lee, Young Ju Cho, Heejun Park

    Purpose: The purpose of this study is to identify any structural change and analyze detailed aspects of the e-commerce market in Korea.Methods: The Cumulative Sum(CUSUM) test was conducted with time-series data including e-commerce transaction value and user data including MAU, usage time, download, etc. from 2015 to 2024. Transaction value data were aggregated solely from transactions made through mobile applications, and user data were collected using a mobile application data analysis platform. Based on the results of the CUSUM test, detailed aspects of user data at the structural change points were analyzed.Results: The results of this study confirmed three structural change points in the last 10 years of Korea’s e-commerce market. The first structural change was confirmed in early 2018. The second structural change was identified in late December 2020, and the third occurred around the end of 2023. The number of active users showed a slightly different pattern compared to the three structural changes, with the number of users growing significantly due to the COVID-19 pandemic and then fluctuating repeatedly. Usage time exhibited somewhat different patterns between the time per user and the time per session. The time per user increased significantly until early 2018 but has been steadily decreasing since then. In contrast, the time per session continuously decreased until early 2018, then showed a slight upward trend immediately after the COVID-19 outbreak but has been steadily decreasing since.Conclusion: The size of Korea's e-commerce market has demonstrated consistent growth over the past decade, undergoing significant structural transformations. The results of this study confirm that the growth of the e-commerce market has been accompanied by structural changes. Notably, while the number of users, downloads, and retention rates fluctuate over time, the usage time exhibits a continuous decrease. It seems that the time users spend on e-commerce applications is getting shorter, which explains the industry’s adoption of various strategies, such as gamification, to increase user’s time spent on applications.

  • Open Access

    A Study on the Effect of Social Cognitive Factors on Employees’Intention to Comply with BYOD Security Policy

    by Chang Geun Lee, Joon Woo Yoo, Jun Sung Park, Young Ju Cho, Jun Young Yoo, So Young Kim, Hee Jun Park

    Purpose: This study focuses on examining the social cognitive factors that influence employees' intentions to comply with security policies in a Bring Your Own Device (BYOD) environment, based on a model derived from Social Cognitive Theory. Additionally, it aims to propose managerial strategies to improve employees' adherence to BYOD security policies.Methods: To investigate employees' intentions to comply with BYOD security policies, we performed hypotheses testing within the proposed research model. A Partial Least Squares Structural Equation Modeling (PLS-SEM) analysis was employed to examine the collected data. The data was gathered from 268 employees working at companies that have adopted BYOD, through an online survey conducted between June 10 and 17, 2024.Results: The results confirmed all of our proposed hypotheses. Encouragement by others, the information security practices by others and instrumental support significantly influence both outcome expectations and self-efficacy. Additionally, instrumental support, outcome expectations, and self-efficacy are shown to affect employees' intentions to comply with BYOD security policies.Conclusion: Our study revealed that companies implementing BYOD should actively encourage employees to minimize security policy breaches by utilizing various support programs and sharing success stories of high performance resulting from adherence to the policies. This approach can effectively increase employees' intention to comply with BYOD information security policies.

  • Top Cited Papers

    How can I trust you if you’re fake? Understanding human-like virtual influencer credibility and the role of textual social cues

    by Joon Woo Yoo, Junsung Park, Heejun Park

    Purpose: This study explores the influence of textual social cues on virtual influencers' perceived attractiveness, homophily and credibility, and their impact on consumers' purchase intentions. The moderating role of perceived anthropomorphism is also assessed.Design/methodology/approach: A randomized between-subjects experiment with 265 participants (134 low social cue/131 high social cue) was conducted. Participants viewed a fictional virtual influencer’s social media profile and post, then completed a survey. Partial least squares structural equation modeling (PLS-SEM) analysis was used to examine the effects of textual social cues on attractiveness, attitude homophily, credibility and purchase intention as well as the moderating role of perceived anthropomorphism.Findings: The study found that textual social cues directly influence attractiveness and attitude homophily, which significantly impact virtual influencer credibility. Credibility, in turn, strongly predicted purchase intention.Practical implications: Incorporating textual social cues into a virtual influencer’s profile to create a likable persona can help overcome the novelty effect and build lasting relationships with followers. Marketers should use textual cues, like emojis and self-disclosure, to enhance marketing effectiveness and select virtual influencers aligned with their target audience.Originality/value: This study is among the first to explore the role of textual social cues in virtual influencers, extending the source credibility model and social information processing theory to the influencer marketing context.


  • Open Access

    Enablers and Inhibitors of Generative AI Usage Intentions in Work Environments

    by JunSung Park, Heejun Park

    Purpose: This study aims to investigate the factors influencing the adoption of Generative AI in the workplace, focusing on both enablers and inhibitors. By employing the dual factor theory, this research examines how knowledge support, customization, entertainment, perceived risk, realistic threat, and identity threat impact the intention to adopt Generative AI technologies such as ChatGPT.Methods: Data were collected from 192 participants via MTurk, all of whom had experience using Generative AI. The survey was conducted in June 2024, and the data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) to ensure the validity and reliability of the measurement model. Attention-check questions were used to ensure data quality, and participants provided demographic information at the end of the survey.Results: : The findings reveal that knowledge support and entertainment significantly enhance the intention to adopt Generative AI, whereas realistic threat poses a substantial barrier. Customization, perceived risk, and identity threat did not significantly affect adoption intentions.Conclusion: This study contributes to the literature by addressing the gap in understanding the adoption mechanisms of Generative AI in professional settings. It highlights the importance of promoting AI's knowledge support and entertainment capabilities while addressing employees' concerns about job security. Organizations should emphasize these benefits and proactively mitigate perceived threats to foster a positive reception of Generative AI technologies. The findings offer practical implications for enhancing user acceptance and provide a foundation for future research in this area.


  • Open Access

    The Impact of Industrial Clusters’ Quality on the Production Resilience in the Global Crisis

    by Young Ju Cho, Chang-Geun Lee, Jun-Young Yoo, So-Young Kim, Hee-Jun Park

    Purpose: This paper aims to verify the difference in production resilience between local clusters and regions without clusters before and after a major crisis. Furthermore, this paper aims to identify the clusters’ quality factors that impact clusters’ shock vulnerability and resilience.Methods: Utilizing open-source data from the US Cluster Mapping platform, this paper compares regions with industrial clusters to those without using the Differences-in-Differences (DID) estimator. It considers the regions with industrial clusters as a treatment group and others as the control group, comparing the period before and after the pandemic. Additionally, the paper examines which cluster factors make a difference in economic resilience during the crisis using Regression Discontinuity Design (RDD).Results: The study finds that regions with industrial clusters show higher production resilience compared to regions without clusters. Moreover, the number of establishments, annual payrolls, and employment can have a positive impact on resilience during the pandemic shock.Conclusion: Though clusters could be vulnerable during the global crisis, industrial clusters can contribute to regional economic development and production resilience in the long-term aspect. Thus, it is required to construct a high-quality local cluster and support it during the economic crisis in the long-term aspect.