Sense adaptive multimodal information fusion: A proposed model

Over the last several years there has been a substantial increase in multimedia contents on the web. Numerous of methods have been proposed so far for effective retrieval of semantic information from multimodal repositories. However, the giant semantic gap is still the bottleneck issue for the researchers of both industry and academia. To overcome this issue there is an utmost need to design such a model that can understand the users information need and retrieve semantically related information from multimedia repository. In literature, three broad categories of multimodal fusion were extensively used viz. early, late and transmedia fusion. In this paper, a study on the working of these fusion approaches has been discussed. Furthermore, a novel hybrid model is proposed that utilizes the benefits of all the three approaches. Thus assumed to minimize the semantic gap issue.