The Cultural Heritage Exploration and Retrieval with Intelligent Systems (CHExRISH) project addresses the fragmentation of heritage data across the Jagiellonian University Archives, Museum, and Library by developing the Jagiellonian University Heritage Metadata Portal (JUHMP)—a Linked Data–based semantic infrastructure aligned with FAIR and open science principles. JUHMP integrates archival, bibliographic, and museum metadata within a shared interoperability framework grounded in the CIDOC‑CRM ontology, enabling coherent representation of people, places, events, and artefacts across institutional boundaries. Conceptually, JUHMP treats academic heritage not as a static collection but as a dynamic network of actors, practices, and intellectual artefacts, aligning with the Sampo model’s evolution from aggregation portals toward analytical and AI-driven discovery environments.

Technically, JUHMP builds a unified integration layer through event-based modelling and authority reconciliation, linking person records from the Corpus Academicum Cracoviense, bibliographic metadata from the Alma system, and museum data from MuzUJ. A hybrid approach combining machine learning–based record linkage and expert validation produced an authority file of 824 verified individuals forming the core of JUHMPv1.2. The system is deployed as a Sampo-based portal, supporting cross-collection search, semantic enrichment, and transparent data exploration.

By operationalising CIDOC‑CRM as an institutional mediation layer, JUHMP demonstrates how research universities can develop semantic heritage infrastructures that enhance interoperability, enable machine-assisted knowledge discovery, and critically reflect on the epistemic foundations of their own data practices.