Test the hidden-information-channel hypothesis for low-status nodes

Investigate whether betweenness analysis identifies a servant-class or other low-status node whose brokerage persists or increases after the Grand View Garden investigation while Wang Xifeng’s centrality declines, using the pre-registered node-permutation null model.

Background

The case study introduces H4b as a falsifiable hypothesis about network mechanisms that may not have been formalized by close reading. It targets servant-class nodes appearing in both the pre-investigation and post-investigation periods and tests whether one of them maintains or increases brokerage after Wang Xifeng’s centrality decreases.

The hypothesis is explicitly unresolved: the figure describes it as open and falsifiable, and the text states that neither a positive nor a null outcome is assumed. The test is scheduled for the full campaign, which is not reported in the paper.

References

The second is a falsifiable hypothesis, H4b: betweenness analysis of low-status nodes may reveal hidden information channels, for instance a servant whose brokerage persists or rises after the investigation even as the mistress's centrality declines. The test is pre-registered: the target set is servant-class nodes appearing in both periods, and the statistic is their maximum brokerage change against a node-permutation null. If the data support H4b, the system surfaces a network mechanism close reading has not formalized; if not, the null result is reported and the case study still demonstrates the evolution-analytics capability. Neither outcome is assumed in advance, and no numbers are reported for the case study in this installment; it is scheduled with the full campaign.

— NS-ST-GraphRAG: Neuro-Symbolic Spatio-Temporal GraphRAG for Literary Knowledge Processing  (2609.05139 - Lin, 4 Sep 2026) in Section 5, subsection “Case Study: Network Evolution around the Grand View Garden Investigation”; Figure 5 caption