HPKS Technical Brief

The Ontology Works High Performance Knowledge Server (HPKS™) is our ontology-driven, scalable, deductive data base management system (DBMS) suitable for use in complex logical domains when fast, accurate answers to complex, challenging questions are needed. It is designed to perform sophisticated analytic queries involving concepts such as but by no means limited to: location and geospatial reasoning, time and temporal reasoning, and provenance and information security. And it does this in semantically and logically complex domains such as intelligence analysis, finance, process control, Enterprise Information Integration, Medical Informatics and many others. The Ontology Works HPKS™ is used in conjunction with our Integrated Ontology Development Environment (IODE™) that allows for the efficient development of complex, high fidelity domain models that are compiled to the HPKS™.

In the IODE™ and HPKS™, Ontology Works combined the expressive power of Common Logic, extended with temporal reasoning and quantification over predicates, with the information storage capabilities of transactional databases to create the first product to manage data in a deductive knowledge-base environment. The result is a dbms that can not only store large amounts of information, but can also meet the most demanding analytic requirements of any business or government agency. The Ontology Works HPKS™ eliminates the costly JOIN tables of relational databases; easily handles ad-hoc querying of data; and leverages the expressive power of Common Logic by storing data according to high-fidelity ontological models that are an order of magnitude more explicit and informative than standard data models. The HPKS™ query tools enable analysts to perform complex, ad hoc queries against the data that are impossible or prohibitively difficult with standard SQL querying. Characteristics of the OW HPKS™ include:

  • Ontology-based deductive DBMS
  • Integrated inference & storage
  • Ontology as schema, views, constraints
  • ACID properties
  • Proof carrying
  • Built-in temporal-interval logic
  • Constraint checking
  • Semantic query optimization
  • Thread-level parallelism
  • Runs on all later versions of Windows and Linux
  • Java 1.5 client-server API

Functionally, the HPKS is comprised of four distinct parts:

Server Presentation Layer

The Server Presentation Layer is Ontology Works' API for programmatically interfacing with all of Ontology Works Knowledge Servers. It is a JDBC-flavored API, consisting of database connection management, logical entity creation, logical fact assertion and deletion, and logical query processing. This layer also provides a number of supporting tools to construct legal queries and logical sentences.

Semantic Optimizer

Ontology Works Semantic Query Optimizer provides a wide range of static query optimizations. Common first order logic queries are manipulated in the context of the ontology the database is using and optimized for parallel execution. The SQO uses many logic transforms to arrive at provably best-case expressions for inputs.

Inference Engine

Ontology Works' in-house inference engine (Pique Engine) is highly optimized for parallel query processing, working with the Semantic Optimizer to allow many parts of a single query expression to run in parallel. This is useful even in a single processor machine since different parts of a query
typically need to access different logical extents, and the parallel nature of the Pique engine means that computational steps can be underway while disk accesses is blocking.

Scalable Storage Layer

The HPKS utilizes Objectivity as its primary storage layer. Objectivity is a high performance Java persistence engine able to handle data in the terabyte range. Ontology Works has layered functionality onto Objectivity to provide features like high performance fact retrieval and ACID properties. Furthermore, Objectivity's disk storage model allows for the Virtual Knowledge Server Architecture whereby multiple HPKS instances all utilize the same Objectivity federation to allow for extremely high rate query access.