HLHoi-Jeong Lim

Hoi-Jeong Lim

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Hoi-Jeong Lim has a total of 43 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    ForensicRAG: A Text-Based Retrieval-Augmented Generation System for Fake News Detection

    by Kelvin Hennry Loudry Malelak, Hoi-Jeong Lim

    The rapid spread of fake news on digital platforms threatens public trust, media integrity, and democratic discourse. Although multiple detection methods exist, such as manual fact-checking, metadata analysis, and linguistic evaluation, these approaches face substantial limitations in managing the increasing volume of misinformation. This study introduces ForensicRAG, an AI-based system that combines retrieval-augmented generation with forensic text analysis to enhance the detection of fake news. The system employs transformer-based classifiers RoBERTa and BERT for binary classification of news articles and integrates TinyLlama, a lightweight generative model, to generate contextual explanations that support classification decisions. A collection of 640 articles was compiled from multiple RSS feeds and subsequently preprocessed and standardized before credibility assessment. The forensic module evaluates linguistic inconsistencies, emotional tone, and semantic markers to strengthen classification reliability. Using a balanced, rigorously curated dataset of 4,000 news articles (2,000 real and 2,000 fake) sourced from Kaggle, RoBERTa achieved 97.8% accuracy, along with strong precision (94%), recall (96%), and F1-score (96%), whereas BERT demonstrated comparably robust performance. These results highlight the potential of ForensicRAG as a scalable, interpretable, and effective tool for combating misinformation in real-world digital environments.

  • Open Access

    Development of a Weighted Average Ensemble Model for Predicting Officially Assessed Land Prices Using Grid Map Data and SHAP

    by Surin Im, Kangmin Kim, Geunhee Lee, Hoi-Jeong Lim

    This study proposes a weighted average ensemble model to predict the Officially Assessed Land Price in Sejong City, South Korea, using 500m × 500m grid-based spatial data. The model analyzes the impact of key variables through SHAP for improved interpretability. Independent variables were grouped into three categories: building characteristics, population demographics, and accessibility to social infrastructure. To construct the predictive model, three boosting algorithms—Gradient Boosting Machine (GBM), eXtreme Gradient Boosting (XGBoost), and LightGBM—were evaluated. The optimal ensemble model, weighted as GBM:XGBoost:LightGBM = 0:9:1, achieved a high predictive performance with R² of 0.8964. SHAP analysis revealed five major factors influencing land price: floor area ratio, accessibility to fire stations, total male population, number of reinforced concrete buildings, and accessibility to general hospitals. The model’s outputs were visualized through a grid-based land price map, revealing elevated land values within Sejong’s Multifunctional Administrative City and around key transportation hubs. By employing fine-scale spatial data and enhancing model interpretability through SHAP, this research presents a robust empirical framework that addresses the limitations of traditional administrative-bound analyses. The findings provide valuable insights for data-driven policy formulation and the evaluation of the real estate market.


  • Open Access

    Meta-analysis in dental research

    by Hoi-Jeong Lim

    Recently, importance of meta-analysis is increasing in the field of dentistry, since it is not easy to settle controversies arising from conflicting studies. Meta-analysis is the statistical method of combining results from two or more individual studies that have been done on the same topic. Merits of meta-analysis includes an increase in power, an improvement in precision, and the ability to address solution not provided by individual studies. However, it might mislead researchers when variation across studies and publication bias are not carefully taken into consideration. The purpose of this study is to help understand meta-analysis by making use of individual results in dental research paper.

  • Open Access

    MIRoBERTa: Mental Illness Text Classification With Transfer Learning on Subreddits

    by Mavin Sao, Hoi-Jeong Lim

    Social media has emerged as a critical resource for text classification, with Reddit prominent among these platforms. In addition to serving as a space for users to share thoughts openly, Reddit is a substantial repository for storing diverse expressions, encompassing discourse on mental health. This study employs the Reddit PullPush application programming interface to gather posts related to seven common mental illnesses: attention-deficit/hyperactivity disorder, anxiety, bipolar disorder, borderline personality disorder, depression, obsessive-compulsive disorder, and post-traumatic stress disorder. A robust collection of 54 161 submissions across 11 subreddits was collected, with the “none” class indicating the absence of mental illness. This study comparatively evaluated three traditional machine learning, two powerful bidirectional deep learning models, and four transformer models: bidirectional encoder representations from transformers (BERT), robustly optimized BERT pretraining approach (RoBERTa), BigBIRD, and longdocument transformer (Longformer) on a multiclass text classification task. The BigBIRD and Longformer approaches outperformed standard transformers, BERT and RoBERTa, on the multiclass text classification task. BigBIRD achieved the highest accuracy and F1-score of 0.840, whereas the RoBERTa model had a slightly lower accuracy and F1-score of 0.834. We pretrain the RoBERTa-base and BERT-base-uncased model using a fill-mask task in the public mental illness corpus domain to improve language understanding before fine-tuning. This pretrained model, mental illness RoBERTa (MIRoBERTa), outperformed other models on the text classification task with an accuracy of 0.847 and F1-score of 0.847. Additionally, mental illness BERT (MIBERT) surpassed existing domain-specific pretrained models with an accuracy of 0.835 and an F1 score of 0.835. We also explore the effectiveness of ensemble techniques by combining the domain adaptation model with the original variant. Finally, we analyze word importance to identify the terms that most significantly contribute to the model classification decisions.


  • Open Access

    Admission Factors and Academic Achievements Associated with the Risk of Grade Retention and National Exam Failure in the School of Dentistry

    by Seungwon Song, Ji-Yeon Jung, Hoi-Jeong Lim

    Objectives: The purpose of this study was to evaluate the risk factors associated with grade retention and national exam failure in the school of dentistry.Methods: The subjects of this study were students who enrolled at the school of dentistry from 2013-2017. Univariable and multivariable analyses were examined through logistic regression analysis. The potential contributing factors of grade retention and national exam failure include: graduated high school, graduated college, major, admission score, admission type, decision type, gender, age at the time of admission, and dental school Grade Point Average(GPA).Results: Higher dental school score correlated with lower risk of grade retention or failing national exam, but undergradu-ate GPA was not related to the risk of national exam failure. Older students and specific admission types had higher risk of repeating school year.Conclusion: It's important to provide focused guidance for high-risk students, especially those with average of dental school score below 80, to prevent grade retention or national exam failure. Developing a student achievement monitoring system can help prevent grade retention or national exam failure.


  • Open Access

    Location Optimization of Emergency Bell Based on Coverage Analysis for Crime Prevention

    by Sun-Woo Lee, Hoi-Jeong Lim, Bo-Gyun Choi

    Typically, emergency bells are security facilities that, when activated, trigger an alarm and immediately dispatch a police car to prevent crime. However, there currently exists an ambiguity in the criteria for emergency bell installation. Consequently, this study aims to find an optimal location for emergency bells whilst considering several factors like cumulative crime incidents. In particular, we exploited emergency bell location data, data on five major crimes, and the geographic information of administrative dongs (primary division of districts) in this study. Specifically, we performed correlation analysis, principal component analysis, and K-means clustering for exploratory data analysis. To effectively cover all 17,437 crimes, which are not covered by the existing emergency bells in Gwangju metropolitan city from 2018 to 2021, the results from the implementation of the emergency bell location set-covering problem revealed the need for about 6228 emergency bells. More precisely, the emergency bell maximal covering location problem was employed to derive the coverage percentage for 250, 500, 800, 1000, and 1500 emergency bells. The results showed that 2850 emergency bells were required to cover over 80% of crime occurrence coordinates, saving over half of the budget compared with covering them all. Overall, this study is noteworthy in its potential role as a roadmap for the optimal placement of emergency bells for future crime prevention.


  • Open Access

    Relationship of academic achievement and residency training according to admission factors in dental school

    by Seungwon Song, Minje Lee, Hoi-Jeong Lim

    Seungwon Song, Minje Lee, Hoi-Jeong Lim. J Korean Acad Oral Health 2022;46:161-8. https://doi.org/10.11149/jkaoh.2022.46.4.161