Search papers, labs, and topics across Lattice.
This paper introduces OMAGR, an ontology-guided framework designed to enhance traffic legal liability determination by addressing the limitations of existing retrieval-augmented generation methods that compress complex legal queries into single pathways. By decomposing queries into ontology-aligned anchors and employing parallel graph retrieval across multiple legal dimensions, OMAGR ensures that interdependent statutory provisions are accurately identified and retrieved. Evaluation on the newly created TrafficLaw-QA dataset reveals that OMAGR significantly outperforms existing baselines in terms of Context Precision and Faithfulness, highlighting its effectiveness in resolving multi-dimensional retrieval challenges.
Parallel multi-anchor retrieval can dramatically improve the accuracy of legal liability determinations in traffic law by overcoming the limitations of traditional single-axis approaches.
Traffic law liability determination is critical for assigning legal penalties, requiring the simultaneous identification of interdependent statutory provisions across multiple legal dimensions. However, existing retrieval-augmented generation methods suffer from a multi-dimensional retrieval bottleneck: single axis architectures compress complex legal queries into a single pathway, causing interdependent statutory dimensions to be overlooked. To address this, we propose OMAGR, an ontology-guided framework that decomposes queries into ontology-aligned anchors and executes parallel graph retrieval across each dimension, ensuring independent retrieval across dimensions before fusion. To evaluate the proposed method, we created the TrafficLaw-QA dataset, an expert-validated benchmark dataset containing 200 questions and 527 legal provisions. Results show that TrafficOmni-RAG outperforms baselines on Context Precision and Faithfulness metrics. The findings demonstrate that parallel multi-anchor retrieval effectively resolves the multi-dimensional retrieval bottleneck, offering a promising direction for traffic law liability determination research.
