RedParrot: Accelerating NL-to-DSL for Business Analytics via Query Semantic Caching
RedParrot is a new framework developed to speed up the conversion of natural language queries into domain-specific languages for business analytics, using a semantic caching mechanism to reduce latency and improve accuracy. By leveraging reusable query skeletons and an entity-agnostic embedding model, it avoids costly LLM pipeline re-computation. The system demonstrates significant performance gains on enterprise data from Xiaohongshu and public benchmarks. It also incorporates a heterogeneous RAG method to handle unseen entities effectively.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Information Retrieval arXiv:2604.22758 (cs) [Submitted on 7 Mar 2026] Title:RedParrot: Accelerating NL-to-DSL for Business Analytics via Query Semantic Caching Authors:Tong Wang, Yongqin Xu, Jianfeng Zhang, Lingxi Cui, Wenqing Wei, Suzhou Chen, Huan Li, Ke Chen, Lidan Shou View a PDF of the paper titled RedParrot: Accelerating NL-to-DSL for Business Analytics via Query Semantic Caching, by Tong Wang and 8 other authors View PDF HTML (experimental) Abstract:Recently, at Xiaohongshu, the rapid expansion of e-commerce and advertising demands real-time business analytics with high accuracy and low latency.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI