AN INFORMATION RETRIEVAL ONTOLOGY BY IMAGES OF HUFI’S STUDENTS

Abstract

Searching for information is an urgent necessity of students in educational institutions in general and Ho Chi Minh City University of Food Industry (HUFI) in particular. Therefore, in this paper, an Ontology describing the HUFI’s image dataset is built semi-automatically to retrieve information by images. To solve this problem, the image dataset is preprocessed and the HOG (Histogram of Oriented Gradient) features are extracted to input for training the SVM (Support Vector Machine) classifier, from there images are classified. Firstly, a hierarchical classes, attributes and relationships of Ontology are created. After that, image individuals are classified and added to Ontology automatically. Finally, the SPARQL query is created from the classification of input image to query on Ontology for extracting the information and similar image. To demonstrate the efficiency of the proposed method, an experiment is conducted and evaluated rely on the Yale Face and images of students from HUFI datasets. The experiment results showed the possibility and efficiency of the method while making it easier to expand the retrieval information related to academics