condition,clarity,argument_logic,contribution_positioning,methodological_rigor,fidelity,usefulness,n_invented_facts,n
A_baseline,5.67,5.04,4.12,4.17,5.83,4.83,0.79,24
B_strong_prompt,5.67,5.17,4.38,4.54,5.21,4.88,1.54,24
C_conventional_rag,5.71,5.42,4.67,4.58,5.29,5.0,1.33,24
D_profiles_retrieval,5.71,5.5,4.46,4.88,5.62,5.04,1.08,24
E_full_engine,5.62,5.96,4.62,6.21,6.25,5.29,0.42,24
E_no_factcheck,5.56,5.81,4.44,5.94,6.12,5.19,0.62,16
E_no_profile,5.75,5.88,4.69,6.12,6.12,5.38,0.69,16
E_no_referee,5.69,5.25,4.31,4.56,5.62,4.88,1.19,16
E_no_retrieval,5.69,5.75,4.62,6.06,6.06,5.31,0.69,16