© 2021 Strange Loop
While the problem of handling massive amounts of data has been at the forefront of database research both in industry and in academia, addressing the complexity of domain models has remained solely a concern of application architects forced to align often highly incompatible problem and solution domains.
HyperGraphDB is a database with a unique memory/data model based on generalized hypergraphs. Those are graphs where edges can point to an arbitrary number of nodes and even to other edges. Thus higher order relationships are expressed naturally which automatically solves most headaches related to domain data modeling. Entities (nodes and edges) have arbitrary values managed by a comprehensive type system embedded as a hypergraph itself.
In a sense, HyperGraphDB is a dynamic-schema database general enough to easily accommodate any meta-model and integrate entities of different formal representations while maintaining high performance through aggressive indexing. In that respect, it is as much a knowledge management system suitable for AI applications as it is database for conventional enterprise systems. Key to such capability are its open-architecture and extremely general formal basis.
In this talk, I will present some of the more interesting aspects of the HyperGraphDB architecture and discuss some of the subtleties in balancing generality, practicality and efficiency in such an open-ended, yet highly organized memory model. I will compare it to other graph databases and put in the larger context of the recent NOSQL movement.