Researchers struggle to manage vast amounts of data coming from hundreds of sources in online repositories. To successfully conduct research studies, researchers need to find, retrieve, filter, extract, integrate, organize, and share information in a timely and high-precision manner. Active conceptual modeling for learning can give researchers the tools they need to perform their tasks in a more efficient, user-friendly, and computer-supported way. The idea is to create "knowledge bundles" (KBs), which are conceptual-model representations of organized information superimposed over a collection of source documents. A "knowledge-bundle builder" (KBB) helps researchers develop KBs in a synergistic and incremental manner and is a manifestation of learning in terms of its semi-automatic construction of KBs. An implemented KBB prototype shows both the feasibility of the idea and the opportunities for further research and development.