Score & visualize community-hub likeness
These methods assume that you understand what your top targets mean within the scope of your research study and broader corpus. For example, if you are studying a rather homogenous and stable group(s) over time, then the top targets can begin to isolate who were important actors being targeted and discussed by these groups, etc. If the groups are heterogenous, then the top targeted maintains the status of who are important actors among these groups, but that doesn't mean that one community hub from period 1 is of similar homiphily as the targets in period 2, for example. Period one's community could be of a conservative partisanship, for instance, while period two's community could be of liberal partisanship.
Analyze and return a list of alike communities across periods based on top targets:
- Init a new
matchingCommunitiesObject
and write a dict of users withmatching_dict_processor()
to send tomatch_maker()
. - Write a list of tuples (matched community pairs and their scores) with
match_maker()
to send tocommunity_grouper()
. - Analyze the intersections and unions of, in this case, the
sorted_filtered_mentions
' values and output a list of sets, where each set includes alike communities across periods in the corpus.
Plot community similarity indices (Jaccard's Co-efficient)
Process and visualize alike communities across periods
- Init a new
allPeriodsObject
and merge all network node and edge data from each period into one dict withbatch_csv()
. - Init new periodObj() as needed with
initializePO()
, retrieve the unique comunity numbers withget_comm_nums()
, and slice up each period per Community into the periodObject withcomm_sender()
. Repeat per Period. - If desired, and as needed, access each periods network data for use.
- Visualize each alike community network graph with
draw_subgraphs()
, a networkx helper function, as desired.