The aim of the project is to determine whether a document is positive, negative or neutral measured by precision performed in the Danish language. Our approach is a further development of the [Li&Liu,2012] approach, which is a cluster based method where the features are uni-grams and negation. Different calculation methods, feature combination and number of clusters are tested. Our result suggests improvement in comparison with the [Li&Liu,2012] cluster with 10-15% for precision. However, our scores are impacted by the quality of the documents used to label the clusters, which is one of the drawbacks of this approach.