Auto-Completion in sourcecode is widely used when it comes to Integrated Development Environments. As soon as the user begins to type the Auto-Completion starts to display recommendations to support the user. Model Recommenders follow similar ideas. They support the user while s/he is modeling and recommend models that may help the user. For examlpe the user can search for a model or a word that occurs in the model and the model recommender will recommend all models which fit the query with an appropriate description and preview. Another approach is to constantly recommend models that may fit to what the user is doing currently, so the context will be analysed. All these techniques are existing. In this thesis we will investigate how it changes the recommended models when not only the current context is analysed but also historic context information. For example the last created or selected classes should be taken into account. A concrete example is that the recommended models will fit the five last working steps which are selection of the class Person, changing of the class name to Human, the movement of the mouse to the class Animal and the selection and deletion of the class Animal.