Novel approach for accurate retrieval of video using semantic annotations

Abstract

The proliferation of Internet has lead to an exponential growth of multimedia content on web. Web contains huge amount of unstructured data, searching for accurate and relevant information is a cumbersome task. Relevant results cannot be fetched from the search engines because they are performing syntactical search for data retrieval. To fetch data efficiently as per the user’s query, semantic search is required. There is a need for a standard way of representing and exchanging information over the internet. In semantic web, data is represented in standard languages like RDF and OWL and linked to the commonly accepted ontologies. As Resource Description Framework (RDF) store the meta information about the data in the form of subject, predicate and object thus, RDF appears to be intuitive solution for making retrieval of videos from web semantic. Semantic annotation of videos increases the efficiency of semantic search by associating the metadata to the resources. In this thesis, the intent is to generate annotated RDF repository of videos for making semantic retrieval using RDF editor. This RDF model is validated through online W3C RDF validation service. SPARQL, RDF query engine, is used to query the validated RDF model to check the efficiency of video search. Ranking of videos has been proposed on the basis of views of a video and rank of pages which have embed video in their web pages; which has been calculated using random web surfer Page rank algorithm.