Agricultural system integrates AI and data to plan soil restoration
Integrating Domain Adaptation and Retrieval-Augmented Generation for Saline-Alkali Land Reclamation Planning
Artificial Intelligence
Summary
Saline-alkali land needs special plans to fix its soil and water problems while staying within budgets. The authors built Wenyao, a tool that reads lots of research papers and uses AI to create customized land reclamation plans. Wenyao combines adapting language models to this topic with searching past cases to improve accuracy. Experts rated Wenyao’s plans highly compared to other AI models, especially when retrieval of past evidence was used. This system helps make science-based plans that professionals can review for better land restoration.
What this means in practice
- •For agricultural planners: Generate land reclamation plans tailored to soil, water, and budget conditions using a system informed by extensive research evidence.
- •For environmental consultants: Support professional review by providing literature-backed reclamation scenarios for saline-alkali soils to inform sustainable land use decisions.
Authors
Zuguang He, Congcong Zheng, Wanyue Chen, Wenya Zhou, Zaijin Yang, Long Long, Yucheng Zhang, Shuqin Gao
Abstract
Reclamation of saline-alkali land requires management plans that account for soil and water conditions, treatment evidence, and financial constraints. This study develops Wenyao, an agricultural decision-support system integrating domain adaptation with retrieval-augmented generation. A bilingual corpus of 8,664 research papers was processed using a 110-indicator schema to construct SA-Bench, comprising 4,860 paper-level planning instances. A QwQ-32B model was adapted through low-rank supervised fine-tuning and combined with case retrieval for budget-conditioned planning. Evidence extraction achieved an F1-score of 0.83 on 300 expert-annotated papers. Without retrieval, Wenyao achieved 95.83% accuracy on 936 source-disjoint multiple-choice questions, compared with 87.93% for the strongest general-purpose baseline; removing explanatory supervision reduced accuracy to 89.42%. Matched-evidence comparisons across 69 region-budget cases supported complementary contributions from adaptation and retrieval. In a blind assessment by 31 experts, Wenyao with retrieval obtained the highest mean scores for objectives and projected outcomes, whereas GPT-5 scored highest for cost estimates. These findings support literature-informed reclamation planning with professional review, while ongoing field monitoring will assess soil and crop responses.