Similarity Measurement For Describe User Images In Social Media

Online social networks like Instagram are places for communication. Also, these media produce rich metadata which are useful for further analysis in many fields including health and cognitive science. Many researchers are using these metadata like hashtags, images, etc. to detect patterns of user activities. However, there are several serious ambiguities like how much reliable are these information. In this paper, we attempt to answer two main questions. Firstly, are image hashtags directly related to image concepts? Can image concepts being predicted using machine learning models? The results of our analysis based on 105000 images on Instagram show that user hashtags are hardly related to image concepts (only 10%of test cases). Second contribution of this paper is showing the suggested pre-trained model predicate image concepts much better (more than 50% of test cases) than user hashtags. Therefore, it is strongly recommended to social media researchers not to rely only on the user hashtags as a label of images or as a signal of information for their study. Alternatively, they can use machine learning methods line deep convolutional neural network model to describe images to extract more related contents. As a proof of concept, some results on food images are studied. We use few similarity measurements to compare result of human and deep convolutional neural network. These analysis is important because food is an important society health field.