Article Per Year (5 Year)
p-Index From 2021 - 2026
2.03
P-Index
This Author published in this journals
All Journal JOURNAL OF INFORMATION SYSTEM RESEARCH (JOSH) Journal of Computer System and Informatics (JoSYC) Jurnal Sistem Komputer dan Informatika (JSON) Bulletin of Computer Science Research KLIK: Kajian Ilmiah Informatika dan Komputer
Muhammad Fikry
Universitas Islam Negeri Sultan Syarif Kasim Riau, Pekanbaru
Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Industrial & Manufacturing Engineering

Published : 14 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 14 Documents
Search

 1   2 
Analisis Sentimen Ulasan Aplikasi Indodax Pada Google Play Store Dengan Algoritma Random Forest Muhammad Iqbal Maulana; Yusra Yusra; Muhammad Fikry; Surya Agustian; Siti Ramadhani
Bulletin of Computer Science Research Vol. 5 No. 4 (2025): June 2025
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletincsr.v5i4.626

Abstract

Crypto assets have become a global phenomenon with a significant increase in the number of investors in Indonesia. Indodax, as the largest crypto asset trading platform in Indonesia, has contributed to the growth of this ecosystem and received many user reviews through the Google Play Store. With more than 5 million downloads and 100 thousand reviews, sentiment analysis is an important tool to understand user perceptions of Indodax services. The results of manual labeling show that the majority of reviews are positive (3989 reviews), while neutral and negative sentiments are 477 and 534 reviews respectively. From the research and testing that has been carried out using the Random Forest method and optimizing with Hyperparameter Tuning GridSearchCV on 4 test scenarios. The best results were obtained in Scenario 4 (3 Preprocessing Stages (Cleaning, Case Folding, and Tokenization) + Random Forest & Hyperparameter Tuning) producing the best value, with Precision 81%, Recall 64%, F1-Score 70% and Accuracy 89%. With the best parameter values ??{'criterion': 'entropy', 'max_depth': None, 'max_features': 'sqrt', 'min_samples_leaf': 1, 'min_samples_split': 2, 'n_estimators': 100}. This study shows that every experimental model that is optimized produces a higher value than experimental model that is not optimized.
Penerapan Metode Support Vector Machine Untuk Analisis Sentimen Pada Komentar Bitcoin Di Aplikasi X Yaskur Bearly Fernandes; Elin Haerani; Fadhilah Syafria; Muhammad Fikry; Lola Oktavia
Bulletin of Computer Science Research Vol. 6 No. 1 (2025): December 2025
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletincsr.v6i1.928

Abstract

Social media has become a primary medium for users to express opinions, including those related to Bitcoin, whose fluctuating value often triggers diverse public responses. The large volume of unstructured comments makes manual sentiment analysis inefficient, thereby necessitating an automated approach based on machine learning. This study aims to classify positive and negative sentiments in Bitcoin-related comments on the X platform using the Support Vector Machine (SVM) algorithm with Term Frequency–Inverse Document Frequency (TF-IDF) feature weighting. The dataset consists of 1,750 Indonesian-language comments labeled by three annotators. The data were processed through several preprocessing stages, including case folding, text cleaning, tokenization, stopword removal, and stemming. Model evaluation was conducted using four data split ratios, namely 90:10, 80:20, 70:30, and 60:40. The experimental results indicate that the 90:10 ratio achieved the best performance, with an accuracy of 72.57%, precision of 0.75, recall of 0.73, and an F1-score of 0.67. The SVM model demonstrates strong performance in identifying positive sentiments; however, it is less effective in detecting negative sentiments due to class imbalance in the dataset. As an additional experiment, testing was performed using a balanced dataset obtained through an undersampling process and several SVM kernel types for comparison. The results show that using a balanced dataset leads to more evenly distributed classification performance across sentiment classes, while the linear kernel provides the most stable performance compared to other kernels. Overall, SVM with TF-IDF weighting proves to be an effective approach for sentiment analysis of Bitcoin-related comments on social media.
Implementasi Metode RBMT dalam Penerjemahan Bahasa Indonesia ke Bahasa Makassar Wan Muhammad Hanif; Yusra Yusra; Muhammad Fikry; Febi Yanto; Siska Kurnia Gusti
Bulletin of Computer Science Research Vol. 6 No. 1 (2025): December 2025
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletincsr.v6i1.935

Abstract

?This research was conducted to address the limited availability of linguistic resources for regional languages, particularly Makassar Language, which does not yet have adequate automatic translation support. The main problem addressed in this study is the absence of a reliable automatic translation system for Makassar Language. The objective of this research is to apply a rule-based translation method to translate text from Indonesian into Makassar Language. This study focuses on the implementation of the Rule-Based Machine Translation (RBMT) method for translating Indonesian text into Makassar Language using the Python programming language. The RBMT implementation involves tokenization, morphological analysis, vocabulary matching, and the application of grammatical rules, including the identification of prefixes and suffixes. The data used consist of a bilingual dictionary compiled from various sources and a set of test sentences representing everyday sentence structures. Translation evaluation was carried out using the Word Error Rate (WER) method, yielding a result of 0.289, and the Character Error Rate (CER) method, with a result of 0.21, which fall into the “Good” category based on the evaluation scale. The main findings indicate that the application of the RBMT method is capable of producing reasonably accurate translations at both the word and character levels. These findings demonstrate that a rule-based approach can be effectively applied to regional languages with limited digital data and provide an initial overview of the potential use of rule-based methods to support the development and preservation of regional languages.
Klasifikasi Hate Speech dan Offensive Language Menggunakan BERT dan Support Vector Machine Muhammad Tirta Syakban; Surya Agustian; Muhammad Fikry; Muhammad Affandes
Bulletin of Computer Science Research Vol. 6 No. 3 (2026): April 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletincsr.v6i3.1061

Abstract

Hate speech and offensive language have become increasingly complex problems on social media, requiring classification approaches that can effectively capture linguistic context. While transformer-based models with end-to-end fine-tuning have become the dominant approach, the use of transformers as fixed feature extractors combined with classical machine learning algorithms remains relatively underexplored, particularly in benchmark settings such as HASOC 2021. This study aims to investigate the effectiveness of a feature-based transformer approach by combining embeddings from BERT and RoBERTa with Support Vector Machine (SVM) classifiers using multiple kernel configurations, including Linear, RBF, Polynomial, and LinearSVC. Experiments were conducted on Sub-task A and Sub-task B by comparing traditional feature-based methods (TF-IDF) with transformer-based embeddings. The experimental results show that RoBERTa embeddings consistently outperform other feature extraction methods. On the test dataset, the combination of RoBERTa and SVM achieves competitive performance compared to other systems in HASOC 2021. In Sub-task B, the optimal model achieves a Macro F1-score of 0.61, outperforming several BERT-based and classical baseline systems.These findings demonstrate that using transformer embeddings as fixed feature representations combined with optimized SVM classifiers can serve as an effective alternative to fine-tuning approaches, particularly in achieving more stable performance under class imbalance conditions. This study contributes by highlighting the potential of feature-based transformer methods as a flexible and competitive strategy for hate speech and offensive language detection.
Co-Authors Abdul Halim Avaldy Rahmat Rivita Dea Ropija Sari Dimas Ferarizki Elin Haerani Elvia Budianita Fadhilah Syafria Febi Yanto Fitri Insani Gusti, Siska Kurnia Iis Afrianty Lola Oktavia Muhammad Affandes Muhammad Iqbal Maulana Muhammad Irsyad Muhammad Riski Muhammad Tirta Syakban Nur Hasanah Hrp Siti Nurhaliza Siti Ramadhani Surya Agustian Syaiful Azhar Wan Muhammad Hanif Yaskur Bearly Fernandes Yusra Yusra Yusra