Abstract Collaborative ?ltering, a widely-used user-centric recommendation technique, predicts an item’s rating by aggregating its ratings from similar users. User similarity is usually calculated by cosine similarity or Pearson correlation coef?cient. However, both of them consider only the direction of rating vectors, and suffer from a range of drawbacks. To solve these issues, we propose a novel Bayesian similarity measure based on the Dirichlet distribution, taking into consideration both the direction and length of rating vectors. Further, our principled method reduces correlation due to chance. Experimental results on six real-world data sets show that our method achieves superior accuracy.