This is my favorite technical review about probabilistic topic models. Topic modeling is an unsupervised ML technique that identifies a set of underlying themes pervading a set of unstructured documents. This review starts by explaining the problem and the basic ideas behind the simplest solution in good detail. It later goes on to discuss contemporary and future research directions to deal with the limitations in existing research. I liked this review because it’s concise, easy to read, and while giving a good overview of the field, it offers fantastic explanations for some of the basic concepts involved along with examples. I found it to be a great primer on the field and an excellent pointer to further reading materials.