Closed BenoitFayolle closed 3 years ago
It was a way on trying to remove clutter from the plot making the plot more crisp as otherwise overlapping clusters is difficult to implement in ggraph. It was a deliberate choice but of course not completely respecting that in a BTM model, one word can be emitted by several topics
Makes sense. I'll edit the code locally and see if my plots get too cluttered as well.
Thanks
Thanks for the package (and BTM), really helpful. I'm getting interesting and useful results.
I notice terms are deduplicated for the biterm clusters graph, hence a term can only appear once in the graph. Is it a feature or just a limitation due to igraph requiring unique vertices names?
https://github.com/bnosac/textplot/blob/d0c40fb84738c0588a4e08784d30eabea26dd52a/R/textplot_biterms.R#L221
I know my topics have words in common and I'd like to see them on the graph if that makes sense?