Assessing Conceptual Similarity to Support Concept Mapping
Proceedings of the Fifteenth International Florida Artificial Intelligence Research Society Conference
Cognition and Multi-Agent Interactions: From Cognitive Modeling to Social Simulation
Cognition and Multi-Agent Interactions: From Cognitive Modeling to Social Simulation
Neural network based constraint satisfaction in ontology mapping
AAAI'08 Proceedings of the 23rd national conference on Artificial intelligence - Volume 2
Feeling and reasoning: a computational model for emotional characters
EPIA'05 Proceedings of the 12th Portuguese conference on Progress in Artificial Intelligence
OWLIM – a pragmatic semantic repository for OWL
WISE'05 Proceedings of the 2005 international conference on Web Information Systems Engineering
Development of intelligent systems and multi-agents systems with amine platform
ICCS'06 Proceedings of the 14th international conference on Conceptual Structures: inspiration and Application
Approximate Spreading Activation for Efficient Knowledge Retrieval from Large Datasets
Proceedings of the 2011 conference on Neural Nets WIRN10: Proceedings of the 20th Italian Workshop on Neural Nets
A cognitive approach to web-based intelligent agents: The TRIPLE architecture
Web Intelligence and Agent Systems
| Hi-index | 0.00 |
This paper introduces a recently proposed hybrid cognitive model, called TRIPLE focusing on its connectionist aspects. They are demonstrated on a series of schematic and realistic examples of analogy based retrieval from memory. The model integrates three modules which run in parallel: serial reasoning, connectionist and emotion engines. The serial reasoning engine deals with the current task, processes perceptual input, plans and performs actions, and synchronizes the activity the other two engines. Its special feature is formal reasoning. This paper focuses on the connectionist engine and the formalism behind it and how it compares to existing approaches. The connectionist engine can be considered as standalone model of retrieval from memory and mapping to the current task. It is based on a series of distributed re-representations of semantic net like symbolic localist representations of knowledge. The distributed representation building mechanisms proposed can reflect taxonomic, relational, semantic or other structure of the long term memory. The representations obtained are made dynamic by a complementary activation spreading process. The mapping-to-task mechanisms of the connectionist module are based on a dynamic similarity evaluation between task and memory content and simultaneous selection of the best mappings by a constraint satisfaction network.