Semantic Similarity Measure with Conceptual Graph-Based Image Annotations

This paper presents a novel approach of the semantic similarity measure that support the image retrieval systems. The approach is composed of five stages: (1) data collection, (2) image annotation, (3) conceptual graph representation, (4) similarity matching, and (5) shows a semantic search result. First stage is collecting the contents into database archive. Label Me tool is used to annotate images. Next stage is representing an image into the conceptual graph. Third stage is finding the similarity matching between the conceptual graph and representative graph. Last stage is showing the set semantic of image results. The results are compared to the classification methods. The experimental results indicate that our proposed approach offers significant performance improvements in the interpretation of semantic images, compared, with the maximum of 88.8% accuracy.