A topic model is a probabilistic generative model for extracting a latent structure from discrete data such as text document. Topic models are successfully used in a wide variety of applications including information retrieval, collaborative filtering and image recognition. In this talk, first I will present basics of topic modelling, such as relations with other probabilistic models, and inference. Then, I will present three applications of topic modelling: social annotation data analysis, multi-scale dynamics analysis, and fashion coordinates recommendation, which I worked on recent years.