Educational videos are one of the best means of imparting knowledge to the users/learners. Videos can convey information in an effective and interesting manner. These videos can be accessed through online or from stored repositories using queries. Search queries play important role in the retrieval. Whenever a user gives an ambiguous query, the search engine may produce irrelevant results. Thus a lot of time is being spent by the users in retrieving the relevant videos. In order to improve the probability of retrieving relevant results, semantic web technologies are applied. This paper aims to extract keywords from the videos and to find the association between the extracted terms. The associated terms are arranged based on their frequency of occurrences. These terms are used to annotate the video automatically, which in turn improves the retrieval of more relevant videos. An ontology is created by experts based on the e-learning video domain. Videos are grouped based on the keywords and on domain ontology, which also helps in enhancing the retrieval results. Videos containing text are only considered for processing. © Universiti Putra Malaysia Press.