Search papers, labs, and topics across Lattice.
This paper introduces HybridRAG-BN, a novel retrieval-augmented framework designed for knowledge-base question answering (KBQA) in Bangla, which integrates hybrid retrieval techniques and a fine-tuned answer verification model. The framework addresses the unique challenges of low-resource languages by combining BM25 and BGE-M3 retrieval methods with a generative model, while also incorporating a robust post-processing stage for fallback answers. Experimental results show that HybridRAG-BN achieves state-of-the-art performance, securing first place in the competition with token-level F1 scores of 0.71654 and 0.72912 on public and private leaderboards, respectively.
Achieving first place in KBQA competitions, HybridRAG-BN demonstrates that effective retrieval and verification can significantly enhance answer accuracy for low-resource languages like Bangla.
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.
