Foundations of a functional approach to knowledge representation.
Artificial Intelligence
HECODES: a framework for HEterogeneous COoperative Distributed Expert Systems
Data & Knowledge Engineering
Enabling technology for knowledge sharing
AI Magazine
The Active Glossary: taking integration seriously
Knowledge Acquisition - Special issue: Current issues in knowledge modeling
Grounding GDMs: a structured case study
International Journal of Human-Computer Studies
KQML as an agent communication language
CIKM '94 Proceedings of the third international conference on Information and knowledge management
Building Large Knowledge-Based Systems; Representation and Inference in the Cyc Project
Building Large Knowledge-Based Systems; Representation and Inference in the Cyc Project
Embedable Problem-Solving Architectures: A Study of Integrating OPS5 with UMass GBB
IEEE Transactions on Knowledge and Data Engineering
Proceedings of the 7th European Workshop on Knowledge Acquisition for Knowledge-Based Systems
Inferring in Lego-land: an Architecture for the Integration of Heterogeneous Inference Modules
AI*IA '93 Proceedings of the Third Congress of the Italian Association for Artificial Intelligence on Advances in Artificial Intelligence
Hi-index | 0.00 |
The Vital-KR allows construction of modular knowledge-based systems, facilitates reuse of kbs modules, provides standard protocols for integrating software or human agents in an application, and supports experimentation in hybrid kbs design.In designing complex knowledge-based applications, a knowledge engineer first identifies the tasks to be tackled and then, if the application is sufficiently complex, assigns these tasks to the different agents--human or software--capable of carrying them out. These cooperative agents can be characterized in terms of the roles they play in problem solving--for example, explanation, classification, translation, or control. Software agents can also be described by their architecture type--conventional or knowledge-based--and by the representation and inference paradigm they employ--rule-based components or frame-based representations, for example. Knowledge-based applications consisting of a number of cooperating software agents that use diverse representation and inference paradigms are referred to as hybrid. Over the past few years, hybrid systems have generated much interest in the AI community, for several reasons:Many current knowledge-based applications are very large and complex, so it is important to reuse preexisting software modules by integrating them in the application. These modules might use alternative representation and inference paradigms; therefore, cooperation mechanisms supporting the interoperability of heterogeneous inference modules are necessary.A widespread agreement in the AI community is that there is no magic bullet of knowledge representation, or as one group of researchers puts it, "There is no single knowledge representation that is best for all problems, nor is there likely to be one." Therefore, researchers are developing environments that offer the user more choice in knowledge representation paradigms. Such environments can provide software support for the development of hybrid knowledge-based applications. For instance, the Cyc system consists of a few dozen inference mechanisms (the heuristic level) integrated by means of a homogeneous representation (the epistemological level).Radical improvements in communication and networking have increased the advantages of distributed applications characterized by multiple software agents residing on multiple remote hosts and using diverse representations.The Vital-KR, a software architecture for the development of hybrid knowledge-based applications, provides the basic substructure for integrating software and human agents in problem solving. The architecture provides implementation-level support in the Vital workbench, a methodology-based workbench covering the whole life-cycle of knowledge-based system (KBS) development, from requirements specification to implementation. The Vital-KR allows construction of modular KBSs, facilitates reuse of KBS modules, provides standard protocols for integrating software or human agents in an application, and supports experimentation in hybrid KBS design.