There are currently over a hundred STEM (Science, Technology, Engineering, and Mathematics) programs in the U.S., aimed at encouraging students to enter a STEM-related career. Of the fields encompassed by STEM, mathematics is the foundation. However, American teens have demonstrated less ability and aptitude for mathematics than the teens in many other countries. Efforts to reduce this ability gap have focused on building a better communal structure within schools, creating more cohesive math curricula, offering smaller classes, encouraging cooperative learning, and helping students overcome math anxiety at school. There is very little published literature on the potential of guided cyber-based self-learning in STEM education, although there are a variety of online STEM learning web sites available for students to access at or after school. In particular, no scientific approach exists to address STEM education challenges in an outside-school setting. The paper briefly unveil an approach to transformatively addressing how mathematics learning, outside of a formal high school setting, could contribute to students' math aptitude. A Web 2.0 system has been deployed, focusing on promoting students' collaborative learning. A big data architecture is applied to addressing voluminous data, which ensures that unstructured Web 2.0 data are always ready for analysis in real time.