Published Paper Details:

Analysing specific tags for images and other sorts of multimedia is quite easy with Instagram. The combination of tags and images are utilized to furnish automatic image annotation (AIA) systems that learn through real time approach. Many unrelated images are accompanied by a multitude of irrelevant hashtags, also known as stop-hashtags, solely for the purpose of attracting clicks and improving their discoverability. Typically, around 20% of hashtags on Instagram are associated with or illustrative to the image's visual content. In this proposal, a novel approach is presented that utilizes the principles of communal intelligence to discovering appropriate hashtags. More specifically, we picturesque the altered version of the renowed algorithm called HITS which was used in community tagging environment that bestows to be a trustworthy and efficient way to create an exemplary and error-free training sets for the retrieval of image based on the content. We have chosen hashtags as a proof-of principles, by utilizing the Figure-eight crowdsourcing platform which enables the gathering of collective knowledge. Bipartite networks were constructed using the crowdsourcing data, with the first kind of nodes representing annotators and the second kind representing hashtags that have picked out. After ranking the annotators based on their effectiveness in the crowd tagging activity, the HITS algorithm was used to choose the relevant hashtags for each image.