The research objective here is to demonstrate as a proof of concept that an emergent distributed agent recommendation system could deliver concept based justifications rather than statistical based justifications from available document sources. We discuss the role of emergence in computational systems and how conceptual networks are realized differently in such systems. We also discuss how journalism is a particularly relevant case study where big data presents itself as a place where recommendation is necessary but also because of big data, a place where people have difficulty and mistrust of systems that try to give them answers. We present test scenarios for rapid prototyped implementation in Python or similar environment. Although not explicitly built on the Starcat framework, the test scenarios rely on features that follow its style of architecture. Finally we discuss results of this proof of concept level of implementation and how it can be advanced with projects that help realize more fully this novel type of recommendation system.