AI-Supported Semantic Reactivation in a Long-Term Blog Corpus: From Chronological Archive to Semantic Long-Term Memory

Since 2008, Jean-Pol Martin has published more than 2,015 blog entries documenting emerging ideas, encounters, pedagogical projects, political observations, theoretical developments, and human–AI dialogues. Together, these texts now constitute a large and internally connected digital corpus. This working paper examines a concrete case of AI-supported semantic reactivation. In response to a question concerning Lanig Garo’s position within Jean-Pol Martin’s semantic field, an AI-supported search retrieved a blog post from 2010 and reconstructed its relationship to political images of the human being, ecological orientation, collective knowledge construction, and Martin’s later theoretical development. The significance of the case lies not merely in retrieving an old webpage. It demonstrates how a present question can reactivate a peripheral historical element and reposition it within a larger intellectual architecture. The article therefore distinguishes a chronological archive from a semantic long-term memory: older texts become accessible through meaning and contextual relevance rather than through prior knowledge of their date, title, or location. The paper explains what this case adds to earlier publications on Jean-Pol Martin’s semantic field and clarifies its function within the growing Zenodo corpus. It argues that the blog serves as a dynamic space in which ideas emerge and accumulate, while Zenodo stabilises, systematises, preserves, and makes selected structures citable. The article concludes with methodological limitations and research questions concerning semantic retrieval, retrospective corpus structuring, digital intellectual legacies, and human–AI cooperation.