Eka Parima Saragih

Eka Parima Saragih


Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
16 September 2025 in METHODIKA: Jurnal Teknik Informatika dan Sistem Informasi
Verified
Journal Article
Open access
PDF file
ANAhmad Fadhil N
Ahmad Fadhil N
ESEka Parima Saragih
Eka Parima Saragih
Real-time sentiment analysis (SA) plays an increasingly vital role in enhancing player experience through emotion-aware game design. By enabling systems such as dynamic difficulty adjustment, adaptive non-playable character (NPC) behavior, and responsive narrative progression, SA allows games to respond intelligently to player emotions. This study investigates the effectiveness of DistilBERT, a lightweight transformer-based language model, for multi-label emotion classification using the GoEmotions dataset, which includes 27 fine-grained emotion categories. The model’s performance was evaluated in terms of classification accuracy and computational efficiency. Experimental results reveal that DistilBERT delivers surprisingly strong performance despite its reduced size, making it a viable candidate for real-time applications in resource-constrained environments. These findings indicate that lightweight transformer models can support emotionally adaptive gameplay without significant trade-offs in latency or accuracy. Future work will focus on integrating DistilBERT into a live game environment to assess its impact on player engagement and real-time system responsiveness.

30 June 2025 in Journal of Electrical Technology UMY
Verified
Journal Article
Open access
PDF file
ESEka Parima Saragih
Eka Parima Saragih
ASAnggraini Dyah Ayu Sekarlangit
Anggraini Dyah Ayu Sekarlangit
FSFaqih Al Suman
Faqih Al Suman
Online misinformation and polarized discussions require better methods for automatically detecting a text's stance. As digital content increases, identifying whether a news article supports, opposes, or is neutral towards its headline is crucial for fighting the spread of false information. This study presents a hybrid model designed for this task. We combine lexical features from Term Frequency-Inverse Document Frequency (TF-IDF), which captures word-level patterns, with contextual semantic information from a pretrained BERT model (bert-base-uncased). The features from both TF-IDF and BERT's [CLS] token were concatenated and used to train a logistic regression classifier. The model was trained and tested on a filtered version of the Fake News Challenge (FNC-1) dataset, with "unrelated" pairs removed to focus on more nuanced stance classification. The final evaluation of this model achieved 83% accuracy with a macro F1-score of 0.68. This model evaluates best in the Neutral stance (F1-score 0.91), but has some difficulty detecting the stance in the Oppositional class (with an F1-score 0.39). The results of this evaluation show that surface level lexical features combined with deep contextual understanding can improve the performance of stance detection.