Researchers have created a benchmark for retrieving relevant Greek statutory articles in legal question answering tasks. The new benchmark, called GreekBarRetrieval, includes 283 bar-exam questions and 6,308 candidate statutory articles. Experiments with various retrieval methods show that dense retrieval outperforms sparse retrieval, but query reformulation using large language models can improve results. The study also introduces a novel ten-round query reformulation loop that further improves performance. --- Why it matters: This matters to researchers in AI because it provides a new benchmark for evaluating the effectiveness of retrieval methods in legal question answering tasks, specifically for Greek statutory articles. Source: https://arxiv.org/abs/2608.18752