UDS is currently being implemented in the Peirce conceptual graphs workbench and is being used as a domain-independent monitor for state-space search domains at a level that is faster than previous implementations designed specifically for those domains.In addition it provides a useful environment for pattern-based machine learning. In particular, conceptual graphs are stored in a relation-based compact form that facilitates matching. These multiple hierarchies support multiple views of the data with advantages over any of the individual methods. All three hierarchies can be stored as 'levels' in the conceptual graphs hierarchy. The data is stored in three partially-ordered hierarchies: a node hierarchy, a relation hierarchy, and a conceptual graphs hierarchy. Foundational to this view is that all data can be viewed as a primitive set of objects and mathematical relations (as sets of tuples) over those objects. This paper gives a data structure (UDS) for supporting database retrieval, inference and machine learning that attempts to unify and extend previous work in relational databases, semantic networks, conceptual graphs, RETE, neural networks and case-based reasoning.

Universal database structure