ProRetrieval：通过可执行程序合成学习编排混合搜索
ProRetrieval: Learning to Orchestrate Hybrid Search via Executable Program Synthesis
浏览论文内容
中文总结 AI 辅助
ProRetrieval 将语言模型作为检索编排器，通过混合 DSL 合成可执行程序，在两个新基准上训练 Qwen3-4B，其 4B 模型在电商、邮件检索任务中优于 GPT-5.5 等主流模型及各类基线。
中文摘要 AI 辅助
现实世界的检索任务常需结合结构化约束与文本、图像的语义意图，并通过任意布尔逻辑进行组合。现有混合检索管道（如 reciprocal rank fusion 或自查询检索器）仅支持固定形式的组合；近期的强化学习检索器将语言模型训练为单一后端的查询生成器，未将异构检索路径的编排纳入其动作空间。本文提出 ProRetrieval，将语言模型重新定义为检索编排器：给定自然语言查询，它会在混合领域特定语言（DSL）中合成可执行程序，该 DSL 交错了针对结构化字段的 SQL 运算符与针对文本、图像的向量检索原语，其中 SQL 本身提供融合异构候选集的逻辑代数。我们使用 GRPO 和 DAPO 训练 Qwen3-4B，并在分层四元奖励机制下进行优化，在基于亚马逊产品和 Enron 邮件构建的两个新基准上评估性能。我们的 4B 模型在电商数据集上的 Hit@1 为 0.81，优于 GPT-5.5 的 0.69；在邮件数据集上的 Hit@1 为 0.91，优于 GPT-5.5 的 0.86，同时也优于 Claude Opus 4.7 及一系列综合的检索、大语言模型增强、结构化查询和基于图的基线方法。代码与数据可在指定链接获取。
英文摘要
Real-world retrieval often composes structured constraints with semantic intents over text and images through arbitrary Boolean logic. Existing hybrid pipelines such as reciprocal rank fusion or self-querying retrievers admit only a fixed form of composition, while recent reinforcement-learning retrievers train the language model as a query generator for a single backend, leaving the orchestration of heterogeneous retrieval paths outside its action space. We propose ProRetrieval, which recasts the language model as a retrieval orchestrator: given a natural-language query, it synthesizes an executable program in a hybrid DSL interleaving SQL operators over structured fields with vector-retrieval primitives over text and images, with SQL itself providing the logical algebra that fuses heterogeneous candidate sets. We train Qwen3-4B with GRPO and DAPO under a hierarchical four-term reward, and evaluate on two new benchmarks built from Amazon products and Enron email. Our 4B model surpasses GPT-5.5 (Hit@1 0.81 vs. 0.69 on e-commerce; 0.91 vs. 0.86 on email) and Claude Opus 4.7 and a comprehensive suite of retrieval, LLM-augmented, structured-query, and graph-based baselines. Code: https://anonymous.4open.science/r/ProRetrieval/; data: https://huggingface.co/datasets/anonymous-7219/ProRetrieval.
