Journals / Advances in Artificial Intelligence Research / 2026 / Cilt: 6

An Intelligent Research Paper Assistant: Integrated Retrieval, Explainable Domain Classification, and Research-Gap Ranking

Araştırma Konularının Önerilmesi, Özetlerin Taslağının Oluşturulması ve Kısaltmaların Üretilmesi için Büyük Dil Modeli Yaklaşımı

Hamza Shahbaz(Government College University Faisalabad)0009-0005-7552-1585
Pages
—
Publication Date
2026-09-17
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Limitations: Results derive from a generated corpus with probable label co-occurrence inflating domain separability; real deployment requires retraining on verified bibliographic data.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Background: The exponential growth of scientific literature creates a practical bottleneck: researchers must manually navigate domain taxonomies, identify research gaps, formulate titles, and bootstrap a methodology before writing begins.
📊
Key Findings & Data:Methods: This study presents an end-to-end Intelligent Research Paper Assistant built over 499,999 scholarly abstracts spanning 10 domains and 25 subdomains, integrating six modules: title suggestion, abstract retrieval, methodology generation, dual-branch domain classification, post-hoc explainability, and research-gap ranking.
💡
Conclusion & Impact:Limitations: Results derive from a generated corpus with probable label co-occurrence inflating domain separability; real deployment requires retraining on verified bibliographic data.
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:large language model · bert · topic suggestion

Abstract

Background: The exponential growth of scientific literature creates a practical bottleneck: researchers must manually navigate domain taxonomies, identify research gaps, formulate titles, and bootstrap a methodology before writing begins. Methods: This study presents an end-to-end Intelligent Research Paper Assistant built over 499,999 scholarly abstracts spanning 10 domains and 25 subdomains, integrating six modules: title suggestion, abstract retrieval, methodology generation, dual-branch domain classification, post-hoc explainability, and research-gap ranking. A deterministic six-stage NLP pipeline feeds a TF-IDF encoder (12,000 features); an incremental SGD linear baseline and a fine-tuned DistilBERT transformer (66.96M parameters) are trained in parallel, with SHAP and LIME providing global and local explainability. Results: Domain classification achieves Accuracy = 1.00 (test-set per-domain accuracy 99.78%) for both models. SHAP and LIME correctly recover domain-discriminative vocabulary, and a composite novelty score identifies Social History and Digital Humanities as the most under-researched subdomains. Limitations: Results derive from a generated corpus with probable label co-occurrence inflating domain separability; real deployment requires retraining on verified bibliographic data.

Özet

Arka Plan: Bilimsel literatürdeki üstel artış, araştırmacıların alan taksonomileri arasında manuel olarak gezinmesini, araştırma boşluklarını belirlemesini, başlık oluşturmasını ve yazım öncesi metodoloji geliştirmesini zorunlu hale getirerek pratik bir darboğaz oluşturmaktadır. Yöntemler: Bu çalışma, 10 alan ve 25 alt alanı kapsayan 499.999 akademik özet üzerinde inşa edilmiş uçtan uca bir Akıllı Araştırma Makalesi Asistanı sunmaktadır. Sistem; başlık önerisi, özet geri getirme, metodoloji üretimi, çift dallı alan sınıflandırması, sonradan açıklanabilirlik ve araştırma boşluğu sıralaması olmak üzere altı modülü entegre etmektedir. Deterministik altı aşamalı bir NLP hattı TF-IDF kodlayıcıya (12.000 özellik) beslenmektedir. Eş zamanlı olarak artımlı SGD tabanlı lineer bir model ve ince ayarlanmış DistilBERT (66.96M parametre) eğitilmiştir. Global ve lokal açıklanabilirlik için SHAP ve LIME yöntemleri kullanılmıştır. Sonuçlar: Alan sınıflandırma görevinde her iki model de %1.00 doğruluk (test seti alan bazında %99.78 doğruluk) elde etmiştir. SHAP ve LIME, alanları ayırt eden kelime dağarcığını doğru şekilde ortaya çıkarmış; bileşik yenilik skoru ise Sosyal Tarih ve Dijital Beşeri Bilimler alanlarını en az araştırılmış alt alanlar olarak belirlemiştir. Sınırlamalar: Sonuçlar, alan ayrılabilirliğini yapay olarak artırabilecek etiket örtüşmesi içeren üretilmiş bir veri kümesine dayanmaktadır. Gerçek dünya uygulamaları için doğrulanmış bibliyografik veriler üzerinde yeniden eğitim gereklidir.

🔖

Cite This Article

Shahbaz, H. (2026). An Intelligent Research Paper Assistant: Integrated Retrieval, Explainable Domain Classification, and Research-Gap Ranking. Advances in Artificial Intelligence Research, Cilt: 6. https://doi.org/10.54569/aair.1975838

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (23)

  1. Ökten M. “Data Analytics in Energy Systems: Applications of Keyword Extraction, Topic Modelling, and Text Classification in Text Mining”, Gazi University Journal of Science Part C: Design and Technology 14(2) (2026) 824-839; doi: 10.29109/gujsc.1904190.
  2. Aydın N, Erdem OA, Tekerek A. “Comparative Analysis of Traditional Machine Learning and Transformer-based Deep Learning Models for Text Classification”, Journal of Polytechnic 28(2) (2025) 445-452; doi: 10.2339/politeknik.1469530.
  3. Okur Hİ, Tohma K, Sertbaş A. “Relational Turkish Text Classification with Graph Neural Networks”, Journal of Polytechnic 28(3) (2025) 867-879; doi: 10.2339/politeknik.1423293.
  4. Soygazi F, Oğuz D. “Medical Text Classification Using Semisupervised Learning and Bert-Based Models”, Mühendislik Bilimleri ve Araştırmaları Dergisi 7(1) (2025) 60-69; doi: 10.46387/bjesr.1597329.
  5. Gusenbauer M, Endermann J, Huber H, Strasser S, Granitzer AN, Ströhle T. “Fine-tuning SciBERT to enable ASJC-based assessments of the disciplinary orientation of research collections”, Scientometrics 131(4) (2026) 2401-2438; doi: 10.1007/s11192-025-05490-0.
  6. Eşidir KA. “DistilBERT ve Geleneksel Makine Öğrenmesi Modelleri ile Duygu Analizi: Performans Karşılaştırması”, International Journal of Pure and Applied Sciences 11(2) (2025) 444-459; doi: 10.29132/ijpas.1604869.
  7. Khan HU, Naz A, Alarfaj FK, Almusallam N. “Identifying artificial intelligence-generated content using the DistilBERT transformer and NLP techniques”, Scientific Reports 15 (2025) 20366; doi: 10.1038/s41598-025-08208-7.
  8. Li Z, Zhu H, Lu Z, Yin M. “Synthetic Data Generation with Large Language Models for Text Classification: Potential and Limitations”, Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (2023) 10443-10461; doi: 10.18653/v1/2023.emnlp-main.647.
  9. Long L, Wang R, Xiao R, Zhao J, Ding X, Chen G, Wang H. “On LLMs-Driven Synthetic Data Generation, Curation, and Evaluation: A Survey”, Findings of the Association for Computational Linguistics: ACL 2024 (2024) 11065-11082; doi: 10.18653/v1/2024.findings-acl.658.
  10. Özdil MS. “Why Large Language Models Can Say but Cannot Imply”, Turkish Academic Research Review 11(1) (2026) 223-238; doi: 10.30622/tarr.1863625.
  11. Asal B, Oyucu S, Doğan F, Polat O, Aksöz A. “Deep Learning and Explainable AI for Email Phishing Classification: A Comparative Study of TabNet, NODE and FT-Transformer Models”, Journal of Polytechnic 29(3) (2026) 1-13; doi: 10.2339/politeknik.1745083.
  12. Zirekgür M, Karakaya B, Avcı D. “Predicting Student Dropout Using Explainable Artificial Intelligence: The Impact of SMOTE-Based Class Balancing Techniques”, Firat University Journal of Experimental and Computational Engineering 5(2) (2026) 574-593; doi: 10.62520/fujece.1887893.
  13. Avcı R, Kayabaşı Koru G, Avcı YZ. “Explainable Phishing Detection Using XGBoost and SHAP: A Comparative Study on Website Content and URL Structure Datasets”, International Journal of Information Systems and Applications 2(1) (2026) 90-111; doi: 10.71340/ubsud.1846061.
  14. Taşar Yildirim T, Gözel N, Çelebi SB, Taşar B, Yaman O, Onay Demir S, Bulu A, Karaduman G. “Explainable AI Framework for Urinary Biomarker-Based Early Detection of Pancreatic Cancer: Clinical Implications of Autoencoder-Optimized XGBoost”, Cerasus Journal of Medicine 3(1) (2026) 37-56; doi: 10.70058/cjm.1831642.
  15. Alp G, Soygazi F. “Benchmarking Machine Learning and Transfer Learning Approaches for Fruit Classification Using Explainable Artificial Intelligence”, Dokuz Eylül University Faculty of Engineering Journal of Science and Engineering 28(83) (2026) 302-310; doi: 10.21205/deufmd.2026288316.
  16. Kelemen A, Szabó ZK, Bozóki S, Szádoczki Z, Hartvig ÁD. “A sensitivity analysis of composite indicators: Min/max thresholds”, Environmental and Sustainability Indicators 23 (2024) 100453; doi: 10.1016/j.indic.2024.100453.
  17. Feldmeyer D, Folch DC, Tate E. “Quantifying the Impact of Sampling Error in Composite Index Construction”, Social Indicators Research 182(1) (2026) 15; doi: 10.1007/s11205-026-03827-7.
  18. Bulut S. “Artificial Intelligence Research in Smart Cities: A Bibliometric Analysis”, Kent Akademisi 19(2) (2026) 1-22; doi: 10.35674/kent.1669862.
  19. Kartal N. “Artificial Intelligence and Machine Learning Applications in Climate Change Research: A Bibliometric Analysis”, International Journal of Advances in Engineering and Pure Sciences 38(2) (2026) 339-350; doi: 10.7240/jeps.1883565.
  20. Turasay FY, Kosar E, Koklu M. “Bibliometric Analysis of Scientific Trends at the Intersection of Artificial Intelligence, Cloud Computing, and Network Technologies”, Karamanoglu Mehmetbey University Journal of Engineering and Natural Sciences 8(1) (2026) 29-38; doi: 10.55213/kmujens.1955728.
  21. Yıldırım N, Yavuz M. “AI-Driven Digital Transformation in Higher Education: A Bibliometric Analysis of Research Trends”, European Journal of Technique 16(1) (2026) 60-68; doi: 10.36222/ejt.1932336.
  22. Akgün M. “Generative Artificial Intelligence in K-12 Education: Bibliometric Analysis and Trends”, İnönü University Journal of the Faculty of Education 27(1) (2026) 249-268; doi: 10.17679/inuefd.1813624.
  23. Kaplan R, Avcı Ü. “Usability of ChatGPT in Human-Computer Interaction and Its Relationship With Artificial Intelligence Attitude”, International Journal of Information Systems and Applications 2(1) (2026) 1-23; doi: 10.71340/ubsud.1869122.