Text-to-SQL has advanced rapidly with large language models, but complex database queries still require reasoning beyond one-shot generation, including multi-step decomposition, execution-based diagnosis, and targeted correction. We present EvoSQL, a co-evolution framework that formulates SQL synthesis as an iterative interaction between a generator and a critic. Moreover, EvoSQL maintains a contextualized candidate memory, verifies SQL candidates with both execution signals and LLM-based critique, and updates its memory through utility-guided aggregation. To strengthen the underlying generator–critic pair, we further introduce a Self-Distillation Policy Optimization (SDPO) fine-tuning stage that injects execution-aware supervision into current coding LLM backbones. Experiments on Spider and BIRD show that EvoSQL consistently improves open-source models over Maj@16 baselines, with particularly large gains on BIRD-Dev, ranging from +1.37% for Qwen3-4B to +9.19% for Qwen2.5-Coder-3B. SDPO initialization further improves selected backbones on Spider-Test and BIRD-Dev. These results suggest that memory-grounded co-evolution is an effective path toward more reliable and generalizable Text-to-SQL systems.

Open-domain tabular question answering (TQA) aims to answer natural language queries over a data lake that contains a large corpus of heterogeneous tables. Unlike closed-domain TQA, this task operates without pre-provided tables, making it particularly challenging. Existing methods usually first retrieve relevant tables and then answer the question based on the retrieved tables. However, it faces challenges in effectively and efficiently preparing tables from the low-quality, heterogeneous data lake and in operationalizing table information. To address these limitations, we present ArrTab (Agentic retrieval and reasoning on Tables) for open-domain TQA. ArrTab leverages MuRe, a self-supervised Multi-table Retriever trained on questions synthesized by translating table-derived SQL queries into natural language via large language models. Based on MuRe, ArrTab proposes a multi-agent system for open-domain TQA that consists of the Table Retriever Agent, the Table Integration Agent, the Information Extraction Agent and the Answer Agent. Furthermore, to enhance accuracy, a lazy execution strategy is adopted where operations are described in natural language but executed only during the information extraction stage, allowing for globally consistent reasoning. Extensive experiments on real-world datasets demonstrate the superiority of ArrTab.

Data preparation is a critical step in enhancing the usability of tabular data and thus boosts downstream data-driven tasks. Traditional methods often face challenges in capturing the intricate relationships within tables and adapting to the tasks involved. Recent advances in Language Models (LMs), especially in Large Language Models (LLMs), offer new opportunities to automate and support tabular data preparation. However, why LMs suit tabular data preparation (i.e., how their capabilities match task demands) and how to use them effectively across phases still remain to be systematically explored. In this survey, we systematically analyze the role of LMs in enhancing tabular data preparation processes, focusing on four core phases: data acquisition, integration, cleaning, and transformation. For each phase, we present an integrated analysis of how LMs can be combined with other components for different preparation tasks, highlight key advancements, and outline prospective pipelines.