# Bridging the Gap: Enabling Natural Language Queries for NoSQL Databases through Text-to-NoSQL Translation Source: [Bridging the Gap: Enabling Natural Language Queries for NoSQL Databases through Text-to-NoSQL Translation]() Collection index: [llms.txt]() Data updated: 2026-06-17T05:00:05.807Z Authors: Jinwei Lu, Jiawei Lu, Chen Zhang, Zhiqian Qin, Haodi Zhang, Yuanfeng Song, Raymond Chi-Wing Wong Published: 2025-02-16T17:01:48.000Z ## Abstract NoSQL databases are core data infrastructure, yet natural-language access to them remains underdeveloped: correct query generation must recover how a non-relational data model represents entities, nested paths, arrays, missing fields, and dynamic keys. This paper studies Text-to-NoSQL, translating natural-language requests into executable NoSQL queries, instantiated with MongoDB aggregation pipelines over schema-less document stores. We present TEND, short for Text-to-NoSQL Dataset, an execution-verified benchmark with 1,210 MongoDB-native tasks across 11 databases. To our knowledge, TEND is the first Text-to-NoSQL benchmark whose database worlds are MongoDB-native by design: experts manually define collection boundaries, nested arrays, optional and sparse paths, polymorphic shapes, and dynamic-key conventions; these worlds are populated with real data and verified through frozen MongoDB execution, so TEND evaluates schema-less document reasoning rather than SQL-to-MQL transfer. We further introduce SAG, a Schema-as-Data Grounding solver that induces path and value grounding from stored-document evidence before bounded MQL generation, execution-grounded repair, and result-consistency selection. Evaluation uses bounded column-tolerant execution accuracy (EXC) as the headline metric, complemented by a graded result-set F1 and a mutually exclusive execution-outcome decomposition. Experiments show that LLMs with strong NL2SQL performance degrade substantially on TEND, validating Text-to-NoSQL as a distinct schema-less document reasoning problem. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Multilingual Table Question Answering on MultiTEND EN](): Multilingual Table Question Answering; MultiTEND EN; EM: 18.85. - [Multilingual Table Question Answering on MultiTEND French](): Multilingual Table Question Answering; MultiTEND French; EM: 18.38. - [Multilingual Table Question Answering on MultiTEND German](): Multilingual Table Question Answering; MultiTEND German; EM (Exact Match): 18.3. - [Multilingual Table Question Answering on MultiTEND Japanese](): Multilingual Table Question Answering; MultiTEND Japanese; EM: 18.05. - [Multilingual Table Question Answering on MultiTEND Russian](): Multilingual Table Question Answering; MultiTEND Russian; EM (Exact Match): 15.89. - [Text-to-NoSQL on MultiTEND Avg of 6 langs](): Text-to-NoSQL; MultiTEND Avg of 6 langs; Exact Match (EM): 17.23. - [Text-to-NoSQL on MultiTEND average of 6 languages (test)](): Text-to-NoSQL; MultiTEND average of 6 languages (test); Exact Match (EM): 17.23. - [Multilingual Table Question Answering on MultiTEND ZH (Chinese)](): Multilingual Table Question Answering; MultiTEND ZH (Chinese); EM: 13.94. ## Original sources and implementation [Original paper]() [PDF]()