English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Zero-shot NL2SQL is crucial in achieving natural language to SQL that isadaptive to new environments (e.g., new databases, new linguistic phenomena orSQL structures) with zero annotated NL2SQL samples from such environments.Existing approaches either fine-tune pre-trained language models (PLMs) basedon annotated data or use prompts to guide fixed large language models (LLMs)such as ChatGPT. PLMs can perform well in schema alignment but struggle toachieve complex reasoning, while LLMs is superior in complex reasoning tasksbut cannot achieve precise schema alignment. In this paper, we propose aZeroNL2SQL framework that combines the complementary advantages of PLMs andLLMs for supporting zero-shot NL2SQL. ZeroNL2SQL first uses PLMs to generate anSQL sketch via schema alignment, then uses LLMs to fill the missing informationvia complex reasoning. Moreover, in order to better align the generated SQLqueries with values in the given database instances, we design a predicatecalibration method to guide the LLM in completing the SQL sketches based on thedatabase instances and select the optimal SQL query via an execution-basedstrategy. Comprehensive experiments show that ZeroNL2SQL can achieve the bestzero-shot NL2SQL performance on real-world benchmarks. Specifically, ZeroNL2SQLoutperforms the state-of-the-art PLM-based methods by 3.2% to 13% and exceedsLLM-based methods by 10% to 20% on execution accuracy.

Interleaving Pre-Trained Language Models and Large Language Models for Zero-Shot NL2SQL Generation