Assessing the relatedness of documents is at the core of many applications such as document retrieval and recommendation. Most similarity approaches operate on word-distribution-based document represe
