|
|
|
@ -14,6 +14,7 @@
|
|
|
|
|
## Only merging lemmas for now, feature selection has no impact on junk classification
|
|
|
|
|
merger <- function(out, text, clean) {
|
|
|
|
|
df <- unnest(out, cols = '_source.ud') %>%
|
|
|
|
|
select(`_id`,lemma,upos) %>%
|
|
|
|
|
unnest(cols = c('lemma','upos')) %>%
|
|
|
|
|
# This line is added in the new merger function, in the old merger function this would result in the following:
|
|
|
|
|
# 1: when using ud, it would result in the string "NA" being present in place of the faulty lemma
|
|
|
|
|