When assigning technological articles as the collaborative writing task, students may nd that the available knowl- edge repositories leave little room for them to contribute and therefore write nothing. To provide guidelines for stu- dents to discover topics, as well as tools to practice problem solving skills, we integrated a computer assisted assessment module into the Mediawiki and employ self-tests as the col- laborative tasks. In these task, item models are used to automatically generate test questions. The items deriving from a same model share a common structure; however, the randomly initialized parameters of the model make them dier from each other. These dierences result in that the answers of an item are usually inapplicable to other items deriving from the same model. Therefore, examinees have to solve these generated items on a case by case basis. Further, how to solve questions deriving from certain models can be served as the topics about which students write articles. The wiki self-assessment system was used in a course on Computer Networks oered to junior students majored in computer science. Five self-test tasks were assigned to 98 students, and they were encouraged to write wiki pages to explain their solution methods. Evidence from this prelim- inary application indicates that the presented approach has a positive eect on learning outcomes.