Garuda - Garba Rujukan Digital

Komparasi Algoritma Machine Learning untuk Deteksi Review Palsu dan Rekomendasi Pembelian Pada Platform Lazada Prabowo, Affan Agung; Barata, Mula Agung; Sa'ida, Ita Aristia
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.307

AbstractThe rapid growth of e-commerce has increased the potential for the emergence of fake reviews that can mislead consumers and reduce the credibility of online purchasing decisions. This study aims to evaluate the performance of several machine learning algorithms in distinguishing fake and genuine reviews, as well as to develop a purchase recommendation model that considers review authenticity. The dataset used consists of 2,644 product reviews from the Lazada platform, which were labeled using a rule-based approach, followed by text preprocessing, normalization, and feature extraction using TF-IDF. The classification methods applied include Support Vector Machine (SVM), Decision Tree, Random Forest, Naive Bayes, and C4.5. The results show that Random Forest and C4.5 achieved the highest accuracy of 99.81%, followed by Decision Tree (99.62%), SVM (98.30%), and Naive Bayes (93.01%). In addition, a purchase recommendation score was developed by combining rating, sentiment, helpfulness, and purchase status to classify products into recommended and not recommended categories. The findings indicate that most reviews identified as fake still result in positive recommendations, which may introduce bias in conventional recommendation systems. Therefore, integrating fake review detection with sentiment analysis and multi-criteria evaluation is essential to improve the reliability of recommendation systems in e-commerce platforms. AbstrakMaraknya perkembangan e-commerce meningkatkan potensi munculnya ulasan palsu yang dapat menyesatkan konsumen dan menurunkan kredibilitas dalam pengambilan keputusan pembelian secara daring. Penelitian ini bertujuan untuk mengevaluasi kinerja beberapa algoritma machine learning dalam membedakan ulasan palsu dan asli, serta mengembangkan model rekomendasi pembelian yang mempertimbangkan keaslian ulasan. Dataset yang digunakan terdiri dari 2.644 ulasan produk pada platform Lazada yang diberi label menggunakan pendekatan rule-based, kemudian melalui tahapan preprocessing teks, normalisasi, dan ekstraksi fitur menggunakan TF-IDF. Metode klasifikasi yang diterapkan meliputi Support Vector Machine, Decision Tree, Random Forest, Naive Bayes, dan C4.5. Hasil pengujian menunjukkan bahwa Random Forest dan C4.5 mencapai akurasi tertinggi sebesar 99,81%, diikuti oleh Decision Tree (99,62%), SVM (98,30%), dan Naive Bayes (93,01%). Selain itu, dikembangkan skor rekomendasi pembelian dengan menggabungkan rating, sentimen, tingkat helpful, dan status pembelian untuk mengelompokkan produk ke dalam kategori direkomendasikan dan tidak direkomendasikan. Temuan menunjukkan bahwa sebagian besar ulasan yang terdeteksi sebagai palsu tetap menghasilkan rekomendasi positif, sehingga berpotensi menimbulkan bias pada sistem rekomendasi konvensional. Oleh karena itu, integrasi deteksi ulasan palsu dengan analisis sentimen serta penilaian multi-kriteria menjadi penting untuk meningkatkan keandalan sistem rekomendasi pada platform e-commerce.

Pengaruh Kombinasi Kapur Dan Fly Ash Terhadap Sifat Mekanik Dan Fisis Tanah Lempung Ekspansif Syaifuddin, Syaifuddin; Mulizar, Mulizar; Cahyani, Agil Ralifa; Reza, Muhammad
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.219

AbstractExpansive clay soils possess high shrink-swell potential, which often leads to structural failures when used as subgrade material. This study aims to evaluate the effectiveness of a combination of lime (KP) and fly ash treatment (FA) in improving the mechanical and physical properties of expansive clay soil obtained from Quarry Cot Kayee Adang, Lhokseumawe City. The research methods included soil classification (AASHTO), physical properties tests, consolidation, compaction (standard Proctor), Atterberg limits, and direct shear tests with additive variations of 0%, 5%, 9%, and 12% lime, along with 5% fly ash. The results showed that the original soil was classified as A-7-6 (36) according to the AASHTO system, with a plasticity index (PI) of 33.80% and a maximum dry density (γdmax) of 1.28 gr/cm³. Adding 9% lime increased the γdmax to 1.32 gr/cm³, and when combined with 5% fly ash, reduced the PI to 22.06%. The direct shear test indicated an increase in cohesion from 0.571 kg/cm² to 0.748 kg/cm² and internal friction angle from 10.7° to 18.1°, resulting in an increase in shear strength (τ) from 7.439 kg/cm² to 12.365 kg/cm². Stabilization using a combination of 9% lime and 5% fly ash proved to be the most effective in enhancing compaction, shear strength, and volumetric stability. These findings recommend the use of this mixture as an efficient alternative for improving expansive clay soil in civil infrastructure construction. Keywords:Expansive clay; fly ash; lime; shear strength; soil stabilization AbstrakTanah lempung ekspansif memiliki potensi kembang susut tinggi yang menyebabkan kerusakan pada konstruksi, terutama jika digunakan sebagai tanah dasar. Penelitian ini bertujuan untuk mengevaluasi efektivitas kombinasi kapur (KP) dan fly ash treatment (FA) dalam memperbaiki sifat mekanik dan fisis tanah lempung ekspansif yang berasal dari Quarry Cot Kayee Adang, Kota Lhokseumawe. Metode penelitian meliputi serangkaian uji laboratorium yang mencakup: uji klasifikasi tanah (AASHTO), uji sifat fisis (kadar air alami, berat volume dan berat jenis), uji batas atterberg, uji konsolidasi,uji pemadatan (standar Proctor), dan uji kuat geser langsung (Direct Shear Test) dengan variasi campuran 0%, 5%, 9%, dan 12% kapur serta 5% fly ash. Hasil pengujian menunjukkan bahwa tanah asli tergolong A-7-6 (36) berdasarkan sistem AASHTO, dengan nilai indeks plastisitas (PI) sebesar 33,80% dan berat volume kering maksimum (γdmak) sebesar 1,28 gr/cm³. Penambahan 9% kapur meningkatkan γdmak menjadi 1,32 gr/cm³ dan menurunkan PI menjadi 22,06% ketika dikombinasikan dengan 5% fly ash. Uji kuat geser menunjukkan peningkatan kohesi dari 0,571 kg/cm² menjadi 0,748 kg/cm² dan sudut geser dalam dari 10,7° menjadi 18,1°, yang berkontribusi terhadap peningkatan kuat geser maksimum (τ) dari 7,439 kg/cm² menjadi 12,365 kg/cm². Stabilisasi menggunakan kombinasi 9% kapur dan 5% fly ash terbukti paling efektif dalam meningkatkan kepadatan, kekuatan geser, dan stabilitas volumetrik tanah. Temuan ini merekomendasikan penggunaan bahan tersebut sebagai alternatif yang efisien untuk memperbaiki tanah lempung ekspansif dalam pekerjaan konstruksi infrastruktur. Keywords:Fly ash; kapur; kuat geser; stabilisasi tanah; tanah lempung ekspansif

Analisis Daya Dukung Tanah Dan Penurunan Pondasi Tiang Pancang Pada Pembangunan Rumah Sakit Arija, Marsal; Abdullah, Zulfhazli; Nanda, Syariaf Asria; Fithra, Herman; Sarana, David; Ariyono, Nabila Attarin; Ruslan, Ruslan
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.306

AbstractFoundations play a critical role in safely transferring structural loads to supporting soil layers. This study aims to evaluate the axial bearing capacity and settlement of pile foundations under both single-pile and pile-group conditions by comparing the Meyerhof method, the Reese Wright method, and numerical analysis using PLAXIS 2D. The analysis is based on Standard Penetration Test (SPT) data and pile foundation design drawings. The results indicate significant variations among the applied methods. The axial bearing capacity of a single pile is estimated at 441.67 tons using the Meyerhof method, 111.669 tons using the Reese Wright method, and 110.64 tons using PLAXIS 2D. The pile group efficiency is calculated as 0.83 using the Converse–Labarre method and 0.727 using the Los Angeles Group method. Meanwhile, the bearing capacity of pile groups is determined to be 293.40 tons, 74.141 tons, and 167.68 tons using the Meyerhof, Reese Wright, and PLAXIS 2D approaches, respectively. The settlement analysis yields values of 25.3 mm using the Vesic method and 37.7 mm from numerical simulation, both of which are within allowable limits according to SNI standards. These findings indicate that the Reese Wright method provides results closer to numerical analysis, making it more representative for foundation design under similar soil conditions.Keywords:Bearing capacity; Meyerhof method; Plaxis; Pile; Standard penetration test. AbstrakPondasi berfungsi sebagai elemen struktur yang menyalurkan beban bangunan ke tanah pendukung secara aman. Penelitian ini bertujuan untuk mengevaluasi daya dukung aksial dan penurunan pondasi tiang pancang pada kondisi tiang tunggal dan kelompok dengan membandingkan metode Meyerhof, Reese Wright, serta analisis numerik menggunakan PLAXIS 2D. Data yang digunakan berupa hasil Standard Penetration Test (SPT) dan gambar rencana pondasi. Hasil analisis menunjukkan perbedaan yang cukup signifikan antar metode. Daya dukung tiang tunggal diperoleh sebesar 441,67 ton dengan metode Meyerhof, 111,669 ton dengan metode Reese Wright, dan 110,64 ton dari PLAXIS 2D. Efisiensi kelompok tiang sebesar 0,83 dengan metode Converse–Labarre dan 0,727 dengan metode Los Angeles Group. Daya dukung kelompok tiang masing-masing sebesar 293,40 ton, 74,141 ton, dan 167,68 ton untuk metode Meyerhof, Reese Wright, dan PLAXIS 2D. Penurunan pondasi sebesar 25,3 mm (Vesic) dan 37,7 mm (PLAXIS 2D), yang masih memenuhi batas aman SNI. Hasil penelitian menunjukkan bahwa metode Reese Wright memberikan estimasi yang lebih mendekati hasil numerik, sehingga lebih representatif untuk perencanaan pondasi pada kondisi tanah serupa. Kata Kunci: Daya dukung; Metode Meyerhof; Plaxis; Standard penetration test; Tiang pancang.

Validating an Automated Essay Evaluation System for Vocational EFL Learners Using Icnale-Gra and a CEFR-J Writing Level Analyzer (CWLA) Mahlil, Mahlil; Masykar, Tanzir; Azhari, Teuku
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.315

AbstractThis paper focuses on the authenticity of an automated essay grader in determining English writing competence in vocational higher learning students in an English as a Foreign Language (EFL) setting. The study was done in Politeknik Negeri Lhokseumawe, Indonesia where first-year second-semester students of the Information and Communication Technology (ICT) Department participated. The primary dataset consisted of 40 student essays that were gathered at the final writing examination. The essays were graded with the use of an AI-based essay grading through the help of an essay grading tool of Writecream (app.writecream.com), which would provide automated writing scores on the basis of the linguistic and structural characteristics. In an attempt to authenticate the validity of the AI scoring results, this study has used two benchmarking frameworks. The International Corpus Network of Asian Learners of English (ICNALE Online) with special attention to the ICNALE-GRA database was utilized as a corpus-based reference, and it represented the standardized levels of writing proficiency of learners. Second, the Writing Level Analyzer (CWLA) based on CEFR was used where the essays were categorized into the bands of the writing proficiency on the CEFR-J framework. The use of CVLA3 desktop version assisted in the classification of the essays into those bands. The validation process was conducted against the AI-generated scores against CEFR-J level classifications and corpus-based proficiency standards. The results showed that the automated essay assessment system showed a validity rate of 62% which showed moderate level of validity with the results of the CEFR-J aligned assessment. This implies that AIs scoring devices can be of service when it comes to vocational EFL formative writing assessment. Nevertheless, the moderate validity raises the importance of using automated systems alongside human judgment particularly in high-stakes measures as with final examination. Altogether, the research adds practice to the existing body of research on AI-aided language testing and underlines the relevance of validation based on CEFR in education. Keywords:AES; CEFR-J; CWLA; ICNALE-GRA corpus; EFL writing assessment. AbstrakPenelitian ini berfokus pada otentisitas penilai esai otomatis dalam menentukan kompetensi menulis bahasa Inggris pada mahasiswa pendidikan tinggi vokasional dalam konteks Bahasa Inggris sebagai Bahasa Asing (EFL). Studi ini dilakukan di Politeknik Negeri Lhokseumawe, Indonesia, di mana mahasiswa semester dua tahun pertama Jurusan Teknologi Informasi dan Komunikasi (TIK) berpartisipasi. Dataset utama terdiri dari 40 esai mahasiswa yang dikumpulkan pada ujian menulis akhir. Esai-esai tersebut dinilai dengan menggunakan penilaian esai berbasis AI melalui bantuan alat penilaian esai Writecream (app.writecream.com), yang akan memberikan skor menulis otomatis berdasarkan karakteristik linguistik dan struktural. Dalam upaya untuk mengotentikasi validitas hasil penilaian AI, studi ini menggunakan dua kerangka kerja benchmarking. International Corpus Network of Asian Learners of English (ICNALE Online) dengan perhatian khusus pada basis data ICNALE-GRA digunakan sebagai referensi berbasis korpus, dan mewakili tingkat kemampuan menulis standar para pembelajar. Kedua, Writing Level Analyzer (CWLA) berbasis CEFR digunakan di mana esai dikategorikan ke dalam tingkatan kemampuan menulis pada kerangka CEFR-J. Penggunaan versi desktop CVLA3 membantu dalam klasifikasi esai ke dalam tingkatan tersebut. Proses validasi dilakukan terhadap skor yang dihasilkan AI terhadap klasifikasi tingkat CEFR-J dan standar kemampuan berbasis korpus. Hasil menunjukkan bahwa sistem penilaian esai otomatis menunjukkan tingkat validitas 62% yang menunjukkan tingkat validitas moderat dengan hasil penilaian yang selaras dengan CEFR-J. Ini menyiratkan bahwa perangkat penilaian AI dapat bermanfaat dalam hal penilaian penulisan formatif EFL kejuruan. Meskipun demikian, validitas moderat meningkatkan pentingnya penggunaan sistem otomatis bersamaan dengan penilaian manusia, khususnya dalam pengukuran berisiko tinggi seperti ujian akhir. Secara keseluruhan, penelitian ini menambah praktik pada penelitian yang ada tentang pengujian bahasa yang dibantu AI dan menggarisbawahi relevansi validasi berbasis CEFR dalam pendidikan..

Analisis Karakteristik Modul PV Monocrystalline dan Polycrystalline di Lingkungan Tropis Basyir, Muhammad; Finawan, Aidi; Dewi, Arsy Febrina; Mauliza, Yuli; Sitorus, M Fitri Anggito
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.324

Monocrystalline and polycrystalline photovoltaic modules continue to dominate solar power system applications; however, experimental evidence based on field measurements in tropical environments remains relatively limited. This study aims to analyze and compare the performance of both PV module types through outdoor testing conducted over 30 days in Indonesia, considering the combined effects of module temperature, dust deposition, and relative humidity. The method employed was comparative field testing with real-time monitoring of key electrical parameters, namely voltage, current, and power, as well as environmental variables. The results indicate that monocrystalline modules consistently outperform polycrystalline modules, with an average power output of 102.4 W (8.6% higher) and a conversion efficiency of 17.8% (11.9% higher). Statistical analysis confirms that the performance difference is significant (p 0.0001; Cohen’s d = 1.85). Module operating temperature was identified as the dominant factor contributing to power degradation, followed by dust and humidity. Collectively, these three environmental factors resulted in an estimated power loss of 13.4% for monocrystalline modules and 16.0% for polycrystalline modules. These findings suggest that monocrystalline modules are better suited for small-scale and residential PV applications in tropical regions due to their superior performance, higher efficiency, and greater thermal resilience.

The Effect Of The Number Of Processes On Multiple Processing Performance On Linux Ekawijana, Ardhian; Noviansyah, Beri
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.288

Processing large data sequentially is often inefficient and time-consuming. Parallel computing is a fundamental solution to accelerate computation by dividing tasks among multiple processing units. This research aims to analyze the influence of the number of processes on the performance of parallel computing implemented using the `fork()` system call on the Linux operating system. A C program was developed to perform a CPU-intensive task on a stock price dataset. Testing was conducted with varying numbers of processes: 1, 2, 4, and 8. The performance metrics measured were wall time, speedup, and efficiency. The test results show a significant reduction in execution time as the number of processes increases. The system achieved near-linear speedup (2.00x for 2 processes, 4.00x for 4 processes, and 7.93x for 8 processes) with high efficiency (99%). These findings prove that the `fork()`-based multi-process approach is highly effective for CPU-bound tasks.

Pengetahuan Guru Vokasi dalam Pembelajaran Berbasis Produksi: Studi Sosio Kulturral pada Immplementasi Teaching Factory di SMK Purwanto, Dedi; Pramono, Agus; Wardoyo, Siswo
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.297

AbstractTeaching Factory (TEFA) is increasingly recognized as a strategic approach in vocational education for linking school learning with authentic industrial practices through production-based learning. However, limited research has examined how TEFA shapes the professional knowledge of vocational teachers. This study aims to explore and characterize the knowledge developed by vocational teachers involved in TEFA implementation in vocational high schools (SMK), using a socio-cultural perspective within the framework of Technical and Vocational Education and Training (TVET). An exploratory qualitative design was employed using a repeated concept mapping method. The study involved 18 experienced vocational teachers from three different fields of expertise who had consistently implemented TEFA. Data analysis distinguished between the content and characteristics of teachers’ knowledge. The knowledge content was categorized into six main components, while its characteristics were examined in terms of concreteness, complexity, specificity, and breadth. The findings indicate that teachers’ knowledge in the TEFA context primarily focuses on occupational competence and the management of production-based learning processes. This knowledge demonstrates high levels of concreteness and complexity due to its close connection with real workplace practices and the integration of pedagogical, occupational, and organizational dimensions. Conversely, knowledge related to students and teacher professional development appears less explicitly articulated. The study concludes that TEFA shapes vocational teachers’ knowledge as a situated and knowledge-intensive professional practice. Keywords:Teaching Factory, Vocational Teacher Knowledge, Technical and Vocational Education and Training (TVET), Socio-Cultural Perspective, Vocational High School. AbstrakTeaching Factory (TEFA) semakin diakui sebagai pendekatan strategis dalam pendidikan vokasi untuk menghubungkan pembelajaran di sekolah dengan praktik industri yang autentik melalui kegiatan pembelajaran berbasis produksi. Namun, penelitian yang mengkaji bagaimana TEFA membentuk pengetahuan profesional guru vokasi masih terbatas. Penelitian ini bertujuan untuk mengeksplorasi dan mengkarakterisasi pengetahuan yang dikembangkan oleh guru vokasi yang terlibat dalam implementasi TEFA di Sekolah Menengah Kejuruan (SMK), dengan menggunakan perspektif sosio-kultural dalam kerangka Technical and Vocational Education and Training (TVET). Penelitian ini menggunakan desain kualitatif eksploratif dengan metode concept mapping berulang. Sebanyak 18 guru vokasi berpengalaman dari tiga bidang keahlian yang secara konsisten menerapkan TEFA terlibat dalam penelitian ini. Analisis data berfokus pada pembedaan antara isi dan karakteristik pengetahuan guru. Isi pengetahuan dikategorikan ke dalam enam komponen utama, sedangkan karakteristiknya dianalisis berdasarkan tingkat kekonkretan, kompleksitas, spesifisitas, dan keluasan. Hasil penelitian menunjukkan bahwa pengetahuan guru dalam konteks TEFA terutama berfokus pada kompetensi okupasional dan pengelolaan proses pembelajaran berbasis produksi. Pengetahuan ini menunjukkan tingkat kekonkretan dan kompleksitas yang tinggi karena berkaitan erat dengan praktik kerja nyata serta mengintegrasikan dimensi pedagogis, okupasional, dan organisasional. Sebaliknya, pengetahuan yang berkaitan dengan peserta didik dan pengembangan profesional guru cenderung kurang terungkap secara eksplisit. Penelitian ini menyimpulkan bahwa TEFA membentuk pengetahuan guru vokasi sebagai praktik profesional yang bersifat kontekstual dan intensif pengetahuan. Kata Kunci: Teaching Factory, Pengetahuan Guru Vokasi, Pendidikan dan Pelatihan Vokasi (TVET), Perspektif Sosio-Kultural, Sekolah Menengah Kejuruan.

Democratizing Climate Intelligence Through Localizez Large Language Models for Education and Governance Kusworo, Zulfikar Aji; Siregar, Widyana Verawaty; Ismail, Baharuddin; Hamdhana, Defry
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.325

AbstractClimate change presents complex challenges in Indonesia, where local governments and communities frequently experience information asymmetry and limited access to expert knowledge, particularly in low-resource and low-connectivity regions. This study aims to develop and evaluate a localized, domain-adapted Large Language Model (LLM) that functions as offline-capable climate knowledge infrastructure for education and local governance in Indonesia. The research method employs a design science research methodology comprising four stages: (1) selection of Qwen3-4B as the base model, (2) curation of an Indonesian climate and energy transition corpus containing approximately 12,400 instruction-response pairs (~38 MB) drawn from national climate policy documents, NDC/RPJMN frameworks, renewable energy guidelines, and educational climate science texts, (3) parameter-efficient fine-tuning using QLoRA with LoRA rank r=16, alpha=32, learning rate 2e-4, 3 epochs, per-device batch size 2 with gradient accumulation 4, and 4-bit NF4 quantization, and (4) offline deployment on consumer-grade hardware with task-oriented evaluation against three baseline models (Qwen3-4B-Thinking, Gemma-3-4B, LLaMa-3.1-8B). The results show that the fine-tuned model (Qwen3-4B-REnewbie v1) achieved a 15.4% perplexity reduction on domain-specific test data and an average qualitative score of 9.3/10 across factual accuracy, reasoning structure, and Bahasa Indonesia language compliance, outperforming all baselines (score range 7.0–8.2). The system operates fully offline on consumer-grade hardware with acceptable inference latency. The conclusion drawn from this study is that localized, resource-efficient LLMs can function as practical climate knowledge infrastructure for vocational education and local governance in Indonesia, aligning with Green AI principles and supporting the democratization of climate intelligence in low-connectivity settings. AbstrakPerubahan iklim menghadirkan tantangan kompleks di Indonesia, khususnya bagi pemerintah daerah dan komunitas lokal yang sering mengalami asimetri informasi dan keterbatasan akses terhadap pengetahuan pakar di wilayah dengan sumber daya dan konektivitas terbatas. Penelitian ini bertujuan mengembangkan dan mengevaluasi Large Language Model (LLM) yang dilokalkan dan diadaptasi ke domain iklim sebagai infrastruktur pengetahuan iklim berbasis offline untuk pendidikan dan tata kelola lokal di Indonesia. Metode penelitian ini menggunakan pendekatan design science research yang meliputi (1) pemilihan Qwen3-4B sebagai base model, (2) kurasi korpus iklim dan transisi energi Indonesia berisi sekitar 12.400 pasangan instruksi-respons (~38 MB) dari dokumen kebijakan iklim nasional, kerangka NDC/RPJMN, panduan energi terbarukan, serta teks ilmiah iklim, (3) parameter-efficient fine-tuning berbasis QLoRA (LoRA rank r=16, alpha=32, learning rate 2e-4, 3 epoch, batch size 2 per perangkat dengan gradient accumulation 4, dan kuantisasi 4-bit NF4), dan (4) deployment offline pada perangkat keras kelas konsumen dengan evaluasi berorientasi tugas terhadap tiga baseline (Qwen3-4B-Thinking, Gemma-3-4B, LLaMa-3.1-8B). Hasil penelitian ini menunjukkan model hasil fine-tuning (Qwen3-4B-REnewbie v1) menghasilkan penurunan perplexity sebesar 15,4% pada data uji domain dan skor kualitatif rata-rata 9,3/10 pada dimensi akurasi faktual, struktur penalaran, dan kepatuhan Bahasa Indonesia, mengungguli seluruh baseline (kisaran 7,0–8,2), serta beroperasi sepenuhnya secara offline pada perangkat konsumen dengan latensi inferensi yang dapat diterima. Kesimpulan yang diperoleh dari penelitian ini adalah LLM yang dilokalkan dan hemat sumber daya dapat berfungsi sebagai infrastruktur pengetahuan iklim yang praktis bagi pendidikan vokasi dan tata kelola lokal di Indonesia, selaras dengan prinsip Green AI dan mendukung demokratisasi kecerdasan iklim di wilayah berkonektivitas terbatas.

Numerical Analysis of Aerodynamic Load Characteristics and Pressure Distribution of NACA 4412 Airfoil in Vertical Wind Turbines Nasution, Arya Rudi; Affandi, Affandi; Damanik, Wawan Septiawan; Siregar, Rahmat Fauzi; Harahap, Jagodang
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.303

AbstractThe performance of a wind turbine is largely determined by its blade design, which typically uses airfoil-shaped cross-sections to optimize aerodynamic efficiency. The NACA 4412 airfoil, known for its strong aerodynamic properties, is commonly employed in both aircraft wings and wind turbine blades. Research on this airfoil investigates its lift and drag coefficients under varying wind speeds (2 m/s to 8 m/s), using SolidWorks 2021 for simulations. Results show that higher wind speeds generally increase lift and drag coefficients, with the highest lift coefficient recorded at 6 m/s. However, at 8 m/s, the lift coefficient decreases, highlighting the sensitivity of aerodynamic performance to speed variations. Additionally, pressure distribution increases with speed, while speed distribution decreases. These findings emphasize the importance of speed variations in influencing the aerodynamic behavior of wind turbine blades. The speed variation on the NACA 4412 airfoil affects the lift coefficient, where the lift coefficient and drag coefficient values will increase at each speed, but at a speed of 8 m/s the lift coefficient and drag coefficient will decrease with a lift coefficient value of 1670314e+06 and a drag coefficient value of 7.857e+07. The highest lift coefficient value occurs at a flow speed of 6 m/s of 3.54769e+04, and the highest drag coefficient value occurs at a flow speed of 6 m/s of 2.606e+06.Keywords: Airfoil NACA 4412; CFD; Speed Variations; Lift Coefficient and Drag. AbstrakKinerja turbin angin sebagian besar ditentukan oleh desain bilahnya, yang biasanya menggunakan penampang berbentuk airfoil untuk mengoptimalkan efisiensi aerodinamis. Airfoil NACA 4412, yang dikenal karena sifat aerodinamisnya yang kuat, umumnya digunakan pada sayap pesawat dan bilah turbin angin. Penelitian pada airfoil ini menyelidiki koefisien gaya angkat dan gaya hambatnya dalam berbagai kecepatan angin (2 m/s hingga 8 m/s), menggunakan SolidWorks 2021 untuk simulasi. Hasil penelitian menunjukkan bahwa kecepatan angin yang lebih tinggi umumnya meningkatkan koefisien gaya angkat dan gaya hambat, dengan koefisien gaya angkat tertinggi tercatat pada 6 m/s. Namun, pada 8 m/s, koefisien gaya angkat menurun, yang menyoroti sensitivitas kinerja aerodinamis terhadap variasi kecepatan. Selain itu, distribusi tekanan meningkat seiring dengan kecepatan, sementara distribusi kecepatan menurun. Temuan ini menekankan pentingnya variasi kecepatan dalam memengaruhi perilaku aerodinamis bilah turbin angin. Variasi kecepatan pada airfoil NACA 4412 mempengaruhi koefisien gaya angkat, dimana nilai koefisien gaya angkat dan koefisien gaya hambat akan bertambah pada setiap kecepatan, namun pada kecepatan 8 m/s koefisien gaya angkat dan koefisien gaya hambat akan berkurang dengan nilai koefisien gaya angkat sebesar 1.670314e+06 dan nilai koefisien gaya hambat sebesar 7.857e+07. Nilai koefisien gaya angkat tertinggi terjadi pada kecepatan aliran 6 m/s sebesar 3.54769e+04, dan nilai koefisien gaya hambat tertinggi terjadi pada kecepatan aliran 6 m/s sebesar 2.606e+06.Kata Kunci : Airfoil NACA 4412; CFD; Variasi Kecepatan; Koefisien Angkat dan Hambatan.

Implementasi SMOTE dan GrideSearchCV untuk Klasifikasi Sentimen Imbalanced pada Isu Reshuffle Kabinet Rofi'i, Muhamad; Vikri, Muhammad Jauhar; Rohmah, Roihatur
VOCATECH: Vocational Education and Technology Journal Vol 8, No 1 (2026): April
Publisher : Akademi Komunitas Negeri Aceh Barat

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.38038/vocatech.v8i1.305

AbstractSocial media has become a rapid barometer of public response to national issues. On X (Twitter), users not only share information but also shape opinions, criticism, and support toward political events. Sentiment analysis is therefore essential for capturing public perceptions explicitly and tracing the dynamics of policy acceptance in digital spaces. This study evaluates the effectiveness of SMOTE and GridSearchCV in improving machine learning performance for imbalanced sentiment classification. It is an applied quantitative study with a comparative computational experimental design. The dataset comprises 6,115 Indonesian-language tweets about a cabinet reshuffle, with (83.5%) negative and 16.5% positive sentiment. Data are preprocessed, labeled using a lexicon-based dictionary, vectorized with TF-IDF, and split into training and test sets. Logistic Regression, Naïve Bayes, SVM, and Random Forest are compared under three scenarios: baseline (no SMOTE), SMOTE, and SMOTE with GridSearchCV for hyperparameter search. Macro F1 is the primary metric, supported by a confusion matrix for per-class evaluation, ensuring minority-class performance is not obscured by high aggregate accuracy. Results show the baseline achieves high accuracy (86.02–89.13%) but low positive recall (21.29–42.57%) and macro F1 (62.83–75.09%), indicating majority-class bias. With SMOTE, positive recall rises (58.42–66.34%) and macro F1 improves (71.50–79.11%) while accuracy remains relatively stable (81.77–88.39%). The best model is Logistic Regression with (88.23%) accuracy, (78.90%) macro F1, and (65.84%) positive recall. In conclusion, SMOTE reduces majority-class bias, GridSearchCV stabilizes performance gains, and macro F1 proves more representative than accuracy for imbalanced sentiment data. AbstrakMedia sosial kini menjadi barometer cepat respon publik terhadap isu nasional. Melalui X/Twitter, masyarakat bukan hanya berbagi informasi, tetapi juga membentuk opini, kritik, dan dukungan atas peristiwa politik. Karena itu, analisis sentimen penting untuk menangkap persepsi publik secara eksplisit dan membaca dinamika penerimaan kebijakan di ruang digital. Penelitian ini bertujuan mengevaluasi efektivitas SMOTE dan GridSearchCV dalam meningkatkan kinerja algoritma machine learning pada klasifikasi sentimen yang tidak seimbang. Studi ini menggunakan metode kuantitatif terapan dengan desain eksperimen komputasional komparatif. Dataset penelitian 6.115 tweet berbahasa Indonesia bertopik reshuffle kabinet dengan distribusi 83,5% negatif dan 16,5% positif. Data dipra-pemrosesan, dilabeli memakai kamus lexicon-based, direpresentasikan dengan TF–IDF, lalu dipisah menjadi data latih dan uji. Klasifikasi diterapkan menggunakan Logistic Regression, Naïve Bayes, SVM, dan Random Forest pada tiga skenario: baseline tanpa SMOTE, SMOTE dengan baseline, serta SMOTE+GridSearchCV untuk pencarian hyperparameter. Macro F1 ditetapkan sebagai metrik utama, didukung confusion matrix untuk evaluasi per kelas. Hasil menunjukkan baseline memberi akurasi tinggi (86,02–89,13%), tetapi recall positif rendah (21,29–42,57%) dan macro F1 (62,83–75,09%), menandakan bias kelas mayoritas. SMOTE meningkatkan recall positif (58,42–66,34%) dan macro F1 (71,50–79,11%) dengan akurasi relatif stabil (81,77–88,39%). Kinerja terbaik dicapai Logistic Regression dengan akurasi (88,23%), macro F1 (78,90%), dan recall positif (65,84%). Simpulan penelitian menunjukkan SMOTE efektif menekan bias kelas mayoritas dan GridSearchCV menstabilkan peningkatan kinerja. Macro F1 terbukti lebih representatif daripada akurasi untuk data sentimen timpang, sehingga pendekatan ini layak diterapkan pada kasus klasifikasi teks tidak seimbang lainnya.

Home Page

OAI Link

Editorial Team

Contact

Reviewer

Google Scholar

Contact Name
Tanzir Masykar

Contact Email
tanzir@aknacehbarat.ac.id

Phone
+6285277752225

Journal Mail Official
vocatech@aknacehbarat.ac.id

Editorial Address
Komplek STTU, Jl. Alue Peunyareng, Ujong Tanoh Darat, Meureubo, West Aceh Regency, Aceh 23681

Location
Kab. aceh barat,

Aceh

INDONESIA

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Home Page

OAI Link

Editorial Team

Contact

Reviewer

Google Scholar

Contact Name Tanzir Masykar

Contact Email tanzir@aknacehbarat.ac.id

Phone +6285277752225

Journal Mail Official vocatech@aknacehbarat.ac.id

Editorial Address Komplek STTU, Jl. Alue Peunyareng, Ujong Tanoh Darat, Meureubo, West Aceh Regency, Aceh 23681

Location Kab. aceh barat, Aceh INDONESIA

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Abstract

Contact Name
Tanzir Masykar

Contact Email
tanzir@aknacehbarat.ac.id

Phone
+6285277752225

Journal Mail Official
vocatech@aknacehbarat.ac.id

Editorial Address
Komplek STTU, Jl. Alue Peunyareng, Ujong Tanoh Darat, Meureubo, West Aceh Regency, Aceh 23681

Location
Kab. aceh barat,

Aceh

INDONESIA