A Domain Ontology for Task Instructions
Knowledge graphs and ontologies represent information in a variety of different applications. One use case, the Intelligence, Surveillance, & Reconnaissance: Mutli-Attribute Task Battery (ISR-MATB), comes from Cognitive Science, where researchers use
- PDF / 967,545 Bytes
- 13 Pages / 439.37 x 666.142 pts Page_size
- 10 Downloads / 208 Views
3
DaSe Lab, Kansas State University, Manhattan, KS, USA {aaroneberhart,coganmshimizu,hitzler}@ksu.edu 2 Air Force Research Laboratory, Wright-Patterson AFB, OH, USA {christopher.stevens.28,christopher.myers.29}@us.af.mil Air Force Research Laboratory, Materials and Manufacturing Directorate, Wright-Patterson AFB, OH, USA [email protected]
Abstract. Knowledge graphs and ontologies represent information in a variety of different applications. One use case, the Intelligence, Surveillance, & Reconnaissance: Mutli-Attribute Task Battery (ISR-MATB), comes from Cognitive Science, where researchers use interdisciplinary methods to understand the mind and cognition. The ISR-MATB is a set of tasks that a cognitive or human agent perform which test visual, auditory, and memory capabilities. An ontology can represent a cognitive agent’s background knowledge of the task it was instructed to perform and act as an interchange format between different Cognitive Agent tasks similar to ISR-MATB. We present several modular patterns for representing ISR-MATB task instructions, as well as a unified diagram that links them together.
1
Introduction
Knowledge graphs facilitate data integration across highly heterogeneous sources in a semantically useful way. Knowledge graphs may be equipped with a schema, frequently an ontology, that combines the associative power of the knowledge graph with the semantics of the ontology. Due to this, they are uniquely suited to support research in cognitive science, where it is often necessary to incorporate information from fields like computer science, psychology, neuroscience, philosophy, and more. Cognitive agents are a sub-field of cognitive science and an application of the more broad study of cognitive architectures. Cognitive architectures, like ACT-R [1] for example, are an approach to understanding intelligent behavior and cognition that grew out of the idea of Unified Theories of Cognition [8]. These systems have their roots in AI production systems and some types use rules-based cognition. Many in Computer Science are familiar with inductive themes from a different type, called Connectionism, due to its historic ties with artificial neural networks. Symbolic cognitive architectures, by contrast, are less widely known outside of cognitive science, and are abstracted and explicit like logic programming. c Springer Nature Switzerland AG 2020 B. Villaz´ on-Terrazas et al. (Eds.): KGSWC 2020, CCIS 1232, pp. 1–13, 2020. https://doi.org/10.1007/978-3-030-65384-2_1
2
A. Eberhart et al.
Both ontologies and cognitive architectures deal with symbolic knowledge. Symbolic cognitive architectures typically focus on the plausibility of knowledge and the way in which that knowledge is translated into human behavior within a specific task. Ontologies offer a set of robust mechanisms for reasoning over complex knowledge bases and could help cognitive architectures adapt to tasks in novel environments. One way the two may be integrated is by leveraging the ontology to reduce the specificity of a cognitiv
Data Loading...
