| Summary: | 碩士 === 長榮大學 === 資訊管理學系碩士班 === 101 === Abstract
Bio-monitoring is a way to show how environment change affects living beings, it require correctly observe and recognition on the target objects. The monitored species includes reptiles, amphibians, insects, and birds, etc. Birds are usually chosen as bio-monitoring targets due to the convenience of observation. Domain experts are able to quickly identify and recognize target objects, however, such knowledge and ability is not easy to impart to beginners.
This work utilizes repertory grid (RG) technique to extract the knowledge from domain experts. Through the interview with experts, the visual features of objects and the knowledge constructs can be derived. This work selected 17 bird species with high similarity, and 31 constructs representing the inherent domain knowledge of the experts are extracted from the interview. The derived repertory grid table represents the knowledge of the experts and may assist beginners to identify the difference between his knowledge constructs and those of experts.
In addition, statistical analysis is conducted on the expert repertory grid constructs tables. The result suggests dividing 17 high similar bird species into 4 groups. The decision tree model also assists in distinguishing bird species with least requirement of constructs. The resulted repertory grid method is implemented as an Android application to allow beginners to learn the knowledge for object identification, and realize the insufficiency of knowledge for better bird identification.
Keywords: repertory grid (RG) technique; object recognition; domain expertise extraction; decision tree
|
|---|
