Claim Missing Document
Check
Articles

Found 32 Documents
Search

Handling Highly Imbalanced Flood Data Using K-Means Clustering in Skyline Query Dominance Testing Vega Purwayoga; Zakwan Gusnadi; Winda Ayu Anggraini
International Journal of Applied Sciences and Smart Technologies Vol. 8 No. 1 (2026): Volume 08, Issue 1, June 2026
Publisher : Faculty of Science and Technology, Universitas Sanata Dharma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.24071/b4693z32

Abstract

Skyline query is a recommendation algorithm used to select objects based on multi-attribute preferences, but a key challenge is that its results can be highly imbalanced, where only a small number of objects meet the preferred criteria. This imbalance reduces the reliability of spatial decision-making, including in flood vulnerability assessment. This study addresses the issue by applying a modified Sort-Filter Skyline method that considers maximum and minimum attribute preferences during sorting. The skyline output shows a strong class imbalance, with only 18 areas identified as flood-prone compared to 1,574 non-flood-prone areas. To mitigate this, K-Means clustering is used as a refinement step. The Elbow and Gap Statistic methods recommend three clusters as optimal, while the Silhouette method suggests eight. Cluster distribution analysis shows that three clusters produce a more balanced representation, with Scheme 1 and Scheme 3 showing better balance ratios and lower variation than Scheme 2. Thus, clustering into three groups helps achieve a more representative mapping of flood-prone areas.
HyRoBERTa: Hybrid Robustly Optimized BERT Approach Model for Sentiment and Sarcasm Detection in Post-Flood Social Media Analysis Siti Yuliyanti; Aveny Septi Asriani; Vega Purwayoga; Zakwan Gusnadi
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 10 No 1 (2026): February 2026
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v10i1.6963

Abstract

The detection problem is a crucial step in sentiment classification because it strengthens the validity and reliability of the model's interpretation of ambiguous text, especially in complex social contexts such as post-disaster public communication. Without this detection, the model is prone to significant classification errors. This study presents a hybrid approach for sentiment analysis with sarcasm detection after a flood disaster by combining the RoBERTa model with sequential deep learning architectures such as GRU, LSTM, and BiLSTM. We used a dataset of 17,520 tweets that were pre-processed using cleaning, normalization, and tokenization. Then, the positive class is further detected to determine whether it is sarcasm. The model was trained using a transformer-based transfer learning method with a combination of hyperparameters: the number of epochs, batch size, dropout rate, and learning rate. The experimental results show that the RoBERTa-GRU model achieved the highest accuracy for sentiment classification at 97. 26%, whereas the RoBERTa-BiLSTM model excels in detecting sarcasm with an accuracy of 98. 74%. RoBERTa-BiLSTM excels in sarcasm detection because it provides a bidirectional sequential mechanism and better long-term memory, effectively leveraging RoBERTa's rich embedding to identify contextual contradictions that are characteristic of sarcasm. Meanwhile, RoBERTa-GRU succeeds in sentiment classification because its architecture is more concise yet effective enough to infer dominant sentiments that have been filtered from the robust representation provided by RoBERTa, making the model more efficient for less complex tasks.