Abstract : Scientific Information Systems (SIS) must move beyond data repositories and closed systems, to allow collaborations among different research disciplines, to include new types of data, to control quality data, and to enable semantic interoperability. Archaeological data include textual information, quantifiable values and measures, sketches, photographies, 3D models, and lots of links among data and historical information sources. DBMS are essential component of SIS nevertheless, existing DBMS including NoSQL DB does not provide enough extensibility and cannot meet all the properties required by a SIS. Our contribution is a multi-paradigm data management system approach that relies on master data and ontology-based annotations. We develop a formal model for ontological-based annotations, we show that this model conforms to a semi-ring algebraic structure and we define a subset of algebraic operator to query annotations. We describe the Burgundy Stone project and we show how our approach is instantiated in a collaborative Web platform that allows researchers to build and publish a corpus.