Development of E-learning sites and their materials make it is necessary to help users finding the desired materials. Context-based search engine will help users for the finding task. However that kind of searching can only be done for learning materials that have been semantically signed or annotated. Annotation is given for the article's content or the article's image within. There are many constraints for manually providing annotations to the learning articles such that automatic metadata or annotation generating method is needed. This paper discusses about annotation generating system with two subsystems: annotation recommender for learning material using contextual analysis and image metadata generator. The methods for contextual analysis are Latent Semantic Analysis (LSA) and WordNet-lexical dictionary USAge. Our experimental results showed that subsystems can be used to generate annotation for articles and images in the articles though we have not done combination of two subsystems.