Contextual Topic Model Based Image Recommendation System

With the incredibly growing amount of image data uploaded and shared via the internet, recommender systems have become an important necessity to ease users´ burden on the information overload. Existing image recommendation systems are designed for discovering the most relevant images with a given query image or short query composed by a few words. However, none of them considers deal with long query, where the query could in any length and potentially contains multiple query topics. To address this problem, we present a contextual topic model based image recommendation system. Compared to using a search engine such as Google Image, our system has the advantage of being able to discern among different topics within a long text query and recommend the most relevant images for each detected topic with semantic "visual words" based relevance.