HALLUCINATION MITIGATION IN LLMs VIA HYBRID RAG + CLASSICAL ML VERIFICATION PIPELINE
“Polytechnic Bulletin” Series “Intellect Innovations Investments” · 30.06.2026
DOI: 10.65599/iii6619 ↗Kurz gefasst
Fachartikel in der Zeitschrift “Polytechnic Bulletin” Series “Intellect Innovations Investments”, online erschienen am 30.06.2026. Verfasst von Shavkat Kholov und 3 weiteren Autor:innen.
Abstract (Originalsprache)
Hallucinations - factually incorrect yet confidently stated outputs - remain a critical challenge in large language models (LLMs). This paper proposes a hybrid pipeline that combines Retrieval-Augmented Generation (RAG) with a classical machinelearning verification module (SVM, logistic regression) for automatic hallucination detection and correction. The proposed architecture is evaluated on a corpus of university academic documents in English and Tajik. Experimental results show that the hybrid approach reduces the hallucination rate by 41 % compared to a plain LLM baseline and by 23 % compared to RAG without verification, achieving an F1-score of 0.90 on the English corpus and 0.85 on the Tajik corpus.
Zitieren
APA
Kholov, S., Bobobekov, K., Abdullozoda, H., & Abdullozoda, H. (2026). HALLUCINATION MITIGATION IN LLMs VIA HYBRID RAG + CLASSICAL ML VERIFICATION PIPELINE. “Polytechnic Bulletin” Series “Intellect Innovations Investments”. https://doi.org/10.65599/iii6619
Vancouver
Kholov S, Bobobekov K, Abdullozoda H, Abdullozoda H. HALLUCINATION MITIGATION IN LLMs VIA HYBRID RAG + CLASSICAL ML VERIFICATION PIPELINE. “Polytechnic Bulletin” Series “Intellect Innovations Investments”. 2026. doi:10.65599/iii6619
BibTeX
@article{Kholov2026,
author = {Kholov, Shavkat and Bobobekov, K.M. and Abdullozoda, Husein and Abdullozoda, Hasan},
title = {HALLUCINATION MITIGATION IN LLMs VIA HYBRID RAG + CLASSICAL ML VERIFICATION PIPELINE},
journal = {“Polytechnic Bulletin” Series “Intellect Innovations Investments”},
year = {2026},
doi = {10.65599/iii6619}
}
EUVIMED