An important research problem for Educational Data Mining is to expedite the cycle of data leading to the analysis of student learning processes and the improvement of support for those processes. For this goal in the context of social interaction in learning, we propose a three-part pipeline that includes data infrastructure, learning process analysis with behavior modeling, and intervention for support. We also describe an application of the pipeline to data from a social learning platform to investigate appropriate goal-setting behavior as a qualification of role models. Students following appropriate goal setters persisted longer in the course, showed increased engagement in hands-on course activities, and were more likely to review previously covered materials as they continued through the course. To foster this beneficial social interaction among students, we propose a social recommender system and show potential for assisting students in interacting with qualified goal setters as role models. We discuss how this generalizable pipeline can be adapted for other support needs in online learning settings. [For the full proceedings, see ED592609.]