New Text-to-NoSQL Benchmark Evaluates Schema-less Document Reasoning

Researchers have developed a new method called Text-to-NoSQL to enable natural language queries for NoSQL databases, specifically focusing on MongoDB. They introduced TEND, a benchmark dataset with 1,210 MongoDB-native tasks designed to evaluate schema-less document reasoning. The study also presents SAG, a solver that grounds path and value information from stored documents before generating and repairing queries. Experiments indicate that even models proficient in Text-to-SQL struggle with this distinct schema-less reasoning problem. AI

RANK_REASON Research paper published on arXiv detailing a new method and benchmark for natural language queries on NoSQL databases. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Jinwei Lu
• MongoDB
• SQL
• Tende

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →