BUILDING AN ONTOLOGY-BASED HERBAL MEDICINE TREE SEARCH SYSTEM

Abstract

Traditional Vietnamese medicine is based on practical experience accumulated over thousands of years. While references to medicinal plants are mainly written in books, it is difficult for non-specialists who want to learn how to use medicinal plants. Many medicinal plants people can confuse species based on common names or species with similar shapes, it is easy to confuse without detailed description of morphology and anatomical characteristics. Moreover, people do not know the uses as well as the diseases that medicinal plants can cure so that is the reason why we will present the research and building an ontology-based herbal medicine tree search system. The use of ontology has been widely applied to technology applications. By representing relationships between concepts defined for a particular field of medicinal plants, it is possible to represent and exchange information between humans and computers. Thanks to that, users can get more information. In addition to using the usual medicinal plant names, users can search for medicinal plants by other criteria such as functions, diseases, taboos and useful parts. Besides, searching by image is also an interesting function. when the user only has pictures of certain plants. The YOLOv5 model is used for image classification of medicinal plant leaves to determine what kind of plant is on the frame. The training dataset was collected from medicinal plants in health centers and medicinal gardens. The test set includes 30 of the 70 medicinal plants prescribed by the Ministry of Health in 2014 and the total number of images is 6000.