Fix #7 in textplot_bitermclusters.default which could unintentionally…

… have behaviour which depends on the order of the data
bnosac · Oct 12, 2021 · 6597170 · 6597170
1 parent d0c40fb
commit 6597170
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 2 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,7 +1,7 @@
 Package: textplot
 Type: Package
 Title: Text Plots
-Version: 0.2.0
+Version: 0.2.1
 Maintainer: Jan Wijffels <[email protected]>
 Authors@R: c(
     person('Jan', 'Wijffels', role = c('aut', 'cre', 'cph'), email = '[email protected]'), 

diff --git a/NEWS.md b/NEWS.md
@@ -1,3 +1,8 @@
+## CHANGES IN textplot VERSION 0.2.1
+
+- Fix bug in textplot_bitermclusters.default (and plot.BTM as it uses textplot_bitermclusters.default) which 
+  unintentionally left the option open that if a biterm which was assigned to several topics it could be shown in one of the 2 topics by chance depending on the order of the biterms data. This could only occur if the most emitted words by each of these 2 topics were the same. See issue #7.
+
 ## CHANGES IN textplot VERSION 0.2.0
 
 - Added textplot_embedding_2d

diff --git a/R/textplot_biterms.R b/R/textplot_biterms.R
@@ -227,7 +227,7 @@ textplot_bitermclusters.default <- function(x, biterms,
   if(!missing(which)){
     biterms <- biterms[biterms$topic %in% which, ]
   }
-  biterms <- biterms[, topic_freq := .N, by = list(term1, term2)]
+  biterms <- biterms[, topic_freq := .N, by = list(term1, term2, topic)]
   biterms <- biterms[, list(best_topic = topic[which.max(topic_freq)], cooc = .N), by = list(term1, term2)]
   # biterms <- biterms[, list(best_topic = utils::head(base::names(base::sort(base::table(topic), decreasing = TRUE)), 1),
   #                           cooc = .N), by = list(term1, term2)]