Summary: | 碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 94 === In 21st century, robots are expected to play more diverse roles to help human beings do more daily tasks. They will not be just seen as mechanical arms in today industry. Multi-functional mechanical pets will produce much market value and make life more fun. In fact, some good design concepts are usually based on existing game robots. Excellent design ideas are always be possible to be found in the existing winner robots. Reusing the design effectively by IT technology is the most exciting and challenging issues in knowledge management domain.
With the help of the Stanford’s Ontology framework, our paper propose a Recommender Agent System to enable the existing design knowledge to be visible though PC’s IE browser and be used effectively with the navigation help. Our agent-based system is started from an existing Content Robot Ontology, regarded as the internal knowledge architecture. The TAO (Topic-Assocation-Occurence) model is then constructed to make clearly robot documents appear as valuable topics of knowledge map. A Topic-Map-based prototype is implemented by the International ISO/IEC 13250 XTM (XML Topic Maps) standard.
In practice, the product design would be started from some simple design concepts. Our recommender agent is proposed here to try to capture the concept query from the interactions with the robot designer. With the classification base of our robot ontology, the search operations can be well navigated and then find out some related documents in few mouse clicks.
Our system can act as a simple Concept Search Engine with the GUI base of two operation modes, Knowledge Vein and Topic Navigator. The recommendation result can be presented by way of Spark Visualization and Tabular Map. In addition, a Topic Map Management Platform is also developed to store knowledge in the relational database in the form of Topic Map.
|