Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal Journal of Scientific Research, Education, and Technology Journal of Systems, Information Technology, and Electronics Engineering
Dhega Febiharsa
Universitas Negeri Malang
Computer Science & IT Economics, Econometrics & Finance Education Electrical & Electronics Engineering Engineering Social Sciences

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Optimasi Algoritma CART untuk Deteksi Kejahatan Siber pada Pesan WhatsApp Berdasarkan Pola Bahasa Fina Sari; Dhega Febiharsa; Kurniawati Kuraniawati
Journal of System, Information Technology and Electronics Engineering Vol 6 No 1 (2026): JUNI 2026
Publisher : Universitas Ivet

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31331/jsitee.v6i1.4790

Abstract

Kejahatan siber pada pesan digital, khususnya penipuan, dapat dikenali melalui karakteristik pola bahasa yang terdapat dalam isi pesan. Penelitian ini bertujuan menerapkan algoritma Classification and Regression Tree (CART) untuk mendeteksi kejahatan siber pada pesan WhatsApp berdasarkan pola bahasa serta memperoleh model dengan kinerja pengujian terbaik. Penelitian menggunakan pendekatan kuantitatif deskriptif dengan jenis eksperimen. Dataset yang digunakan berjumlah 1.143 pesan teks dari repositori publik GitHub SMS Spam Classification Bahasa yang digunakan sebagai representasi pesan WhatsApp, terdiri atas 914 data training dan 229 data testing. Data melalui tahap preprocessing yang meliputi case folding, cleaning, tokenizing, dan stopword removal, kemudian dianalisis berdasarkan kata indikator untuk menghasilkan tiga fitur numerik, yaitu normal, promosi, dan penipuan. Algoritma CART dioptimasi menggunakan GridSearchCV dengan 5-Fold Cross Validation melalui beberapa percobaan konfigurasi parameter. Hasil pengujian menunjukkan konfigurasi terbaik menggunakan criterion = gini, max_depth = 5, min_samples_split = 2, dan min_samples_leaf = 3, dengan Accuracy Cross Validation sebesar 81,95% dan Accuracy Testing sebesar 75,55%. Nilai tersebut meningkat sebesar 2,19 poin persentase dibandingkan Accuracy Testing sebelum optimasi sebesar 73,36%. Berdasarkan hasil tersebut, konfigurasi dengan Accuracy Testing tertinggi dipilih sebagai model akhir. Hasil penelitian menunjukkan bahwa optimasi parameter dan analisis pola bahasa dapat mendukung penerapan CART dalam klasifikasi pesan ke dalam kategori Normal, Penipuan, dan Promosi
Digital and AI-Enabled Scaffolding in RPL Assessment: A Systematic Literature Review of Mechanisms, Evidence, and Design Principles Fuad Abdillah; Dhega Febiharsa
Journal of Scientific Research, Education, and Technology (JSRET) Vol. 5 No. 4 (2026): Vol. 5 No. 4 2026
Publisher : Kirana Publisher (KNPub)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.58526/jsret.v5i4.1268

Abstract

This study aims to synthesize the mechanisms, evidence, and principles of digital scaffolding design and artificial intelligence assistance in the assessment of Recognition of Prior Learning (RPL). The study employs a systematic literature review design, adhering to PRISMA 2020 reporting guidelines, with a publication deadline from January 2021 to September 2026. The initial synthesis of the selected literature combines thematic analysis of support mechanisms and structured analysis of evaluation results, distinguishing empirical evidence from conceptual propositions. Four mechanisms were identified: articulation of lessons learned from experience, portfolio development and evidence mapping, formative feedback, and interpretive support for assessors. The literature documented candidate readiness, narrative traceability, and document processing performance, but did not provide adequate evidence regarding improved validity and fairness of recognition decisions. Classification accuracy and annotation consistency measured outcomes that differed from the adequacy of competency evidence. The proposed originality lies in the integration of mechanisms, assessment tasks, evaluation results, and application conditions within a single analytical framework. The study contributes by clarifying the differences between presentation quality, evidentiary quality, and decision validity, and by formulating design principles regarding claim ownership, traceability, assistance adjustment, uncertainty, accessibility, and human accountability. Unfinished corpus selection and quality assessments limit the determination of effectiveness and generalization. This framework provides a basis for testing candidate and assessor support in a variety of RPL contexts.
Co-Authors Fina Sari Fuad Abdillah Kurniawati Kuraniawati