Event

November 29-December 1, 2016 Austin, Texas

Defense Innovation Summit 2016

Autonomous Systems

Our patent-pending semantic equivalence method integrates observational data from multiple sources (e.g., sensor data, textual descriptions) at various levels of abstraction, resolves ambiguity and detects conflicting observations prior to resolving to labeled ontology concept identifiers suitable for reasoning.

Charles Parker and George Garrity will be attending the Defense Innovation Summit this year. We will be presenting an overview of our recent work on poster 313, “Knowledge Extraction from Mixed-Precision Information”, during Poster Session I Tuesday afternoon from 2:30pm-3:15pm. We are actively seeking commercial partners to bring this technology to market.

A fundamental barrier to effective human-machine communication is the lack of a shared, unambiguous language that is understandable to humans and precise enough for machine reasoning. The knowledge of domain experts is aggregated from a variety of information sources, ranging from raw text or data to structured and normalized databases (Mixed Precision Information; MPI).

We introduce a novel standards-based method for extracting knowledge from MPI to provide knowledge workers and machine reasoners with verifiable interpretations of observational data.

Our approach combines semantic and semiotic methods to represent information at multiple levels in concept hierarchies, “slice” and aggregate concepts to represent information consistently for ambiguous human language and reasoners, provide multiple entry points for information (term, concept, data), provide attachment points for reasoning over rules and axioms and accommodate multiple interpretations of information.

Parker et al., Knowledge Extraction from Mixed-Precision Information

Download Abstract (81kB PDF) Download Poster (5MB PDF)

[permalink] Posted November 29, 2016.