This research aims to design an AI web application for office document summarization and question answering using Streamlit and Ollama. The system integrates three models: qwen2.5:7b for RAG chatbot, mistral-nemo:latest for summarization, and nomic-embed-text for embedding. Testing on 15 documents (PDF, DOCX, TXT) with 52 queries showed cosine similarity of 0.87 and precision@3 of 0.92. Caching improved processing speed by 85.5% and RAG reduced hallucination from 42% to 8%. Average ROUGE score was 0.68 and user satisfaction reached 4.3 out of 5. The study concludes that local language models effectively support office document comprehension while maintaining data privacy.

Memuat Halaman 1...
Evaluasi berbasis sistem terhadap data telaah substantif dan metodologis naskah
Komputasi sistem berdasarkan kriteria editorial baku jurnal.
Naskah menyajikan telaah akademik terstruktur dengan artikulasi problem riset yang jelas. Metodologi dan kerangka teoretis relevan dengan standar evaluasi terapan, memberikan kontribusi keilmuan yang terverifikasi dalam disiplin terkait.
System-generated synthesis derived from the completed peer-review and editorial assessment process
This research aims to design an AI web application for office document summarization and question answering using Streamlit and Ollama. The system integrates three models: qwen2. 5:7b for RAG chatbot, mistral-nemo:latest for summarization, and nomic-embed-text for embedding.
Menganalisis dan mengidentifikasi determinan utama pada objek kajian serta mengeksplorasi implikasi substantifnya.
Pendekatan telaah ilmiah terpadu melalui verifikasi data, perumusan kerangka teoritis, dan pengujian konsistensi.
Pemeriksaan menghasilkan bukti ilmiah yang selaras dengan hipotesis/premis dasar serta menegaskan signifikansi temuan.
Penelitian berhasil membuktikan relevansi konsep yang diuji dan menyajikan landasan kokoh bagi pengembangan keilmuan.
Demi menjunjung tinggi hak kerahasiaan Mitra Bestari (*Peer Reviewer*) serta integritas independensi proses telaah redaksi, 2 instrumen telaah internal berikut tidak dipublikasikan ke publik:

Tri-Source Research Quality Assessment • APS-AT-RQS-a4b563c2-v1.0
🟢 Primary: Analytical evidence and high statistical regression model robustness
🟢 Secondary: Practical utility and direct organizational/policy applicability
🟡 Single-institution population context with cross-sectional observation scope.
Official Scholarly Passport & Academic Identity Record · ASIA Index International Database
| Record Field | Official Record Value |
|---|---|
| ASIA Index Record ID | ASIA-2026-729047 |
| Index Status | VERIFIED & INDEXED |
| Record Type | Scholarly Article |
| Publication Origin | APASIFIC Scholarly Ecosystem |
| Date Submit | 26 Agustus 2026 |
| Date Published | 2 September 2026 |
| Record Version | 1.0 (Canonical Release) |
| Field | Details |
|---|---|
| Article Title | PEMANFAATAN MODEL OLLAMA UNTUK APLIKASI SUMMARISASI DAN TANYA JAWAB DOKUMEN PERKANTORAN |
| DOI | 10.5281/zenodo.22242068 |
| Publication Venue / Journal | AJITE - Ilmu Komputer & Teknologi Informasi |
| ISSN / eISSN | Dalam Antrean |
| Subject Category | Computer Science & Information Technology |
| Document Type / Language | Research Article · English |
| Identity Element | Verification Details |
|---|---|
| Authors | Iqbal Firdaus, S.Kom., M.Kom., Gusti Aditya Aromatica Firdaus, S.Kom., M.Kom., Yeni Agus Nurhuda, S.Si., M.Cs., Jordan Abraham |
| ORCID Registry | ✓ Linked / Verified System ID |
| Affiliations | Institut Bisnis dan Teknologi Kalimantan |
| Author Identity Status | ✓ Verified & Provenance Resolved |
| Source / Infrastructure | Status | Record / Evidence |
|---|---|---|
APASIFIC(30/30 pts) | ✓ Origin Verified | Internal Publication Record |
DOI / Zenodo(25/25 pts) | ✓ Verified | 10.5281/zenodo.22242068 |
Zenodo(15/15 pts) | ✓ Linked | Zenodo ID: 22242068 |
OpenAIRE(15/15 pts) | ✓ Discovered | European Research Graph Record |
ORCID(0/10 pts) | Linked | Author Research Identity |
Google Scholar(5/5 pts) | ✓ Discoverable | Scholarly Discovery & Citation Observation |
| Article Metric | Value |
|---|---|
| Total CitationsNon-Self: 0 · Author Self: 0 · Journal Self: 0 | 0 |
| ASIA Article Score (AAS) | 72 |
| Citation Velocity | 1.2 / Year |
| Citation Network Status | Verified Citation Graph |
| Scholarly Chain Score | 90 / 100 |
| Metric Status | ACTIVE |
| Journal Metric | Value |
|---|---|
| ASIA Citation Score (ACS)Corpus Network Density: 0.0158 | 8.9 |
| ASIA Scholarly Rank (ASR)Prestige-weighted Network Metric | 1.98 |
| ASIA Impact Factor (AIF) | 2.85 |
| ASIA Percentile | 92th Percentile |
| ASIA Metric Quartile | AM-Q1 |
| Category Rank | 9 / 100 |
This record represents the indexed scholarly identity, publication provenance, metadata connectivity, and metric status of this article within the ASIA Index ecosystem.
ASIA-CANONICAL-RECORD
DOI
10.5281/zenodo.22242068
VOL 1
EDISI 4
SEP
2026
Evaluasi Kualitas Naskah Akademik
"How often is the research cited?"
Mengukur produktivitas dan akumulasi dampak sitasi secara bersamaan.
"How many works achieved meaningful citation impact?"
Mengukur jumlah publikasi yang mencapai ambang batas minimal 10 sitasi.
"How robust is the research quality supporting the evidence?"
Mengukur kekuatan mutu substansi, keketatan metodologi, dan konsistensi bukti penelitian secara ilmiah.
Iqbal Firdaus*1, Gusti Aditya Aromatica Firdaus2, Yeni Agus Nurhuda3, Jordan Abraham4. (2 Se). PEMANFAATAN MODEL OLLAMA UNTUK APLIKASI SUMMARISASI DAN TANYA JAWAB DOKUMEN PERKANTORAN. AJITE - Ilmu Komputer & Teknologi Informasi, 1(4). https://doi.org/10.5281/zenodo.22242068
Spesifikasi perhitungan metrik kanonikal artikel, jurnal, dan penentuan Quartile ASIA Index.
Alur evaluasi ilmiah: Formula → Score → Percentile → Quartile
Status pengindeksan dan registrasi resmi dengan basis data akademik internasional.

Permanent Legacy Research Identity Record • Standard v1.0
Bagikan langsung ke Facebook, WhatsApp, atau buat kartu Story 9:16 untuk TikTok @apasificacademician.