ABSTRACT
Information explosion in this era has led to the proliferation of digital data in form of image, text, video and audio etc. This information exists in divergent modalities (image, text, audio and video) therefore we are in need of multimodal systems where various modalities could be fused together to get accurate on time information. Uncertainty is a major issue in information access and retrieval models, and incomplete information needs to be treated in Multimodal Systems because imprecision indicates the existence of a value that cannot be measured. There is no denial of fact that uncertainty puts hindrance in obtaining information in real-time systems and as per knowledge rarely any study of multimodal information access and retrieval treats imprecise and inconsistent information inherited in the modelling stage. As different modalities could be represented in neutrosophic domain using neutrosophic sets and theories, this work proposes to use a generalised version of Intuitionistic Fuzzy Sets i.e. Neutrosophic Fuzzy Sets which is a Soft Computing Technique for the treatment of uncertainty which persists in information recovery in Multimodal Systems. Here we have taken only two modalities i.e. image and text to show how we can incorporate the notion of indeterminacy and uncertainty in information retrieval models using image and tag from Flickr dataset. In terms of precision and recall the proposed architecture seems to yield good results.
