2

I am conducting a network text analysis as part of a systematic literature review of social network analysis in computer-supported learning. My nodes are terms in the included studies, and the ties represent how close these terms are to each other in the text (window size of 10). This is my first time conducting any sort of network analysis and I am unsure about interpreting my results.

I ran a weighted clique percolation (k=5) on my data in the hopes of finding subcommunities of terms that could give me a quantitative overview of the content of my pool of studies. Instead of finding overlapping subcommunities, I found one large subcommunity subsuming smaller ones (k=5). What can I make out of this data then? Is it a result still work noting?

with small enough threshold or with small enough k this is what you get for all connected graphs, so this is quite uninformative. I believe you would have to experiment with thresholds and k until you have more than one component, which is arguably a bit ad hoc. Because you have weights, other methods (for example, see, https://stats.stackexchange.com/questions/2717/clustering-with-a-distance-matrix) might work better.

– Valentas – 2017-09-24T08:44:48.203