Quantifying Visual-Representativeness of Social Image Tags Using Image Tag Clarity

Tags associated with images in various social media sharing web sites are valuable information source for superior image retrieval experiences. Due to the nature of tagging, many tags associated with images are not visually descriptive. In this chapter, we propose Image Tag Clarity to evaluate the effectiveness of a tag in describing the visual content of its annotated images, which is also known as the image tag visual-representativeness. It is measured by computing the zero-mean normalized distance between the tag language model estimated from the images annotated by the tag and the collection language model. The tag/collection language models are derived from the bag of visual-word local content features of the images. The visual-representative tags that are commonly used to annotate visually similar images are given high tag clarity scores. Evaluated on a large real-world dataset containing more than 269K images and their associated tags, we show that the image tag clarity score can effectively identify the visual-representative tags from all tags contributed by users. Based on the tag clarity scores, we have made a few interesting observations that could be used to support many tag-based applications.