Speaker
Description
In large-scale scientific research, experimental data faces high acquisition costs and a shortage of high-quality data, while a significant amount of critical data is scattered in unstructured forms across various scientific literature. To address this issue, this study proposes an artificial intelligence framework for constructing high-quality knowledge bases from literature corpora and its intelligent agent applications. First, large models and prompt engineering are used to extract high-precision data from texts, images, and tables in domain-specific scientific literature, which are then transformed into standardized, structured data through a data management system and integrated with experimental data to construct an "AI-Ready" domain knowledge base. Based on this, a research intelligent agent is designed and implemented, capable of converting scientific queries into program instructions via a natural language interface, supporting data retrieval, statistical analysis, and scientific plotting functions. The practical case of building a shale neutron scattering database demonstrates that this framework significantly enhances the efficiency of knowledge base construction and scientific analysis. The framework exhibits strong versatility and is expected to be applicable to more scientific fields in the future, aiding the intelligent transformation of scientific research.