This thesis researches automated services for professionals aiming at starting collaborative learning projects in open learning environments, such as MOOCs. It investigates the theoretical backgrounds of team formation for collaborative learning. Based on the outcomes, a model is developed describing the process of 1) project proposal assessment for fit to learing materials, and 2) project team formation based on prior knowledge and personality. Algorithms for the formation of learning, creative, and productive teams are described. A large scale experiment demonstrates the sucessfull use of Latent Semantic Analysis for the modeling of a knowledge domain, the assessment of project fit to the learning environment, and for the assessment of learner prior knowledge. The outcomes contribute to MOOC design, team formation theory, and the LSA knowledge base.