This work proposes HybridRAG-BN, a retrieval-augmented framework for Bangla KBQA that integrates hybrid retrieval using BM25 and BGE-M3, answer generation using the GGUF version of Gemma-4-31B-Instruct, and a LoRA-fine-tuned Gemma-4-31B-Instruct model for answer verification and refinement.
Abstract
Knowledge-base question answering (KBQA) systems rely on effective retrieval and reasoning mechanisms to generate accurate answers from external knowledge sources. However, developing reliable KBQA systems for low-resource languages such as Bangla remains challenging due to limited retrieval-focused research, scarce language resources, and difficulties in grounding generated responses in external knowledge. In this work, we propose HybridRAG-BN, a retrieval-augmented framework for Bangla KBQA that integrates hybrid retrieval using BM25 and BGE-M3, answer generation using the GGUF version of Gemma-4-31B-Instruct, and a LoRA-fine-tuned Gemma-4-31B-Instruct model for answer verification and refinement. To further improve robustness, the framework incorporates a post-processing stage that addresses unresolved cases through fallback answer replacement and DuckDuckGo-assisted retrieval. Experimental results demonstrate the effectiveness of the proposed framework, achieving token-level F1 scores of 0.71654 and 0.72912 on the public and private leaderboards, respectively, securing first place in the competition.
Evaluations on standard KBQA benchmarks show that the proposed ARI-KBQA enhances model performance with a reduced search space, especially in complex multi-hop query scenarios.
Jian-Qi Gao, Hang Yu, Jian Cao et al.· 0 citations
Arabic question-answering systems in high-sensitivity domains require not only accurate retrieval but also reliable evidence grounding and effective hallucination mitigation, as incorrect or unsupported responses may have serious consequences. Existing retrieval and generation approaches do not fully integrate reliable lexical retrieval, semantic understanding, and evidence-based answer verification within a unified framework for these domains. To address this limitation, this study proposes an Enhanced Hybrid Retrieval-Augmented model that combines BM25-based lexical retrieval, dense semantic scoring, semi-structured metadata, domain-aware classification, source-trust and freshness indicators, and evidence-based answer verification. The model was evaluated on a corpus of 128,297 Arabic documents using 200 expert-validated questions and compared against BM25, Dense, and Classical Hybrid retrieval configurations. Expert assessment showed that the Enhanced Hybrid Model produced 198 fully grounded correct answers out of 200, achieving a grounded-correctness rate of 99.0%, compared with 96.5% for Classical Hybrid, 87.0% for BM25, and 86.5% for Dense. It also achieved the strongest answer-level Hit@5 performance, the best golden-answer ordering, and the highest dynamic mixture-based F1-score. These findings demonstrate that integrating reliable retrieval, metadata-aware ranking, and evidence-grounded answer verification can substantially improve the reliability and trustworthiness of Arabic question-answering systems in high-sensitivity domains, providing a practical foundation for future evidence-based intelligent information systems.
A. Aloqla, Reda Salama, Wajdi Alghamdi et al.· Future Internet· 0 citations
Evaluating linguistically diverse descriptive answers in a consistent and accurate manner in modern digital education systems is a growing challenge, especially in low-resource languages like Hindi. Traditional lexical and rule-based grading systems cannot adequately reflect the meaning behind the words, negation, paraphrasing, and so on, which leads to low grading reliability. To overcome these limitations, this study proposes an automated evaluation framework with intelligent rule-based linguistic preprocessing and transformer-based deep learning. The framework uses a fine-tuned multilingual BERT (mBERT) model bhavikardeshna/multilingual-bert-base-cased-hindi to provide contextual embeddings, and cosine similarity-based semantic alignment with the model l3cube-pune/hindi-sentence-similarity-sbert is used to provide automated scores. With optimal setting of learning rate = 5×10⁻⁴, batch size = 24 and epochs = 40, accuracy, precision, recall and F1 score of 78.9%, 80.6%, 77.4% and 79.0% respectively is achieved on HindiRC-Data-master dataset (24 passages, 127 question-answer pairs, grades 2-5) which is more than 14% higher than lexical similarity baselines and is better than previous Hindi QA architectures without domain-specific preprocessing pipelines. The suggested system will save about 40% manual grading, and will enable scalable, consistent and repeatable assessment.
Unknown authors· international journal of eng...· 0 citations
GCA-KBQA is proposed, a step-wise fine-tuned LLM-based framework that employs hop-wise generation, knowledge-assisted calibration, and path-level assembly to construct complete LFs for KBQA, demonstrating its superior performance compared to state-of-the-art baselines.
Ran-Ran Bu, Jian Cao, Jian-Qi Gao et al.· Annual International ACM SIG...· 0 citations
