Traditional archival management and services take documents as basic information unit. While laying the foundation for China's archival endeavors, this approach suffers from fragmented workflows, misaligned demands, insufficient support, and performance bottlenecks, making it difficult to meet the need for precise archival utilization in the AI era effectively. This paper aims to address the above practical challenges that archival document management faces. From the perspective of granular semantic data representation of archival facts, it proposes the academic concept of archival text fact particle deconstruction, constructs a fact-logic model, explores relevant technical methods, and validates their effectiveness through case studies. Archival text fact particle deconstruction is a technical process carried out by archival institutions, based on preprocessed texts and guided by the fact-logic model, identify relevant entities, attributes, and relationships, and represent them by RDF (Resource Description Framework) format, then transforms factual semantic information of archival document into granular, quasi-isomorphic set of record factors data. Its application is expected to drive the intelligent transformation of China's archival sector and holistically reshape the information ecology of traditional archival document services.