I have a corpus sliced in three different periods.
I performed the network mapping choosing the ‘custom period’ parameters (custom, three periods, regular), and obtained the three maps for each of them.
I also built a sub-corpus (with a query) for one of the three periods. I mapped that sub-corpus with the same parameters as the overall corpus, but the map for the period that i considered differs than the one of the same period that i produced on the overall map. I saw in the log file that the detected threshold differs, as well as the louvain modularity.
But I cannot understand what is the reason behind it, in the mapping procedure, and also wondering what is methodologically more correct, to separate the three periods of my overall corpus in sub-corpus and map them separately, or map the entire corpus selecting the three custom periods.