Find words that are similar in following rows.

find_similar_words.R

library(stringr)
library(dplyr)
library(stringdist)

df <- tibble(word = c("Agrilus pilosivittatus",
                      "Agrilus pilosovittatus", 
                      "foo", 
                      "bar", 
                      "baz", 
                      "something else", 
                      "some stuff", 
                      "some stff"))

# create a lag variable (this "shifts" the whole variable)
df <- df %>% 
  mutate(word_before = lag(word)) %>% 
  mutate(row_num = row_number()) 

# look at the new data (see the "shift"?)
df

# calculate the levensthein distance between the two 
df <- df %>% 
  mutate(dist = stringdist::stringdist(word, word_before))

df

# decide which rows to keep based on levensthein distance
rows_to_keep <- df %>% 
  filter(dist < 2) %>% # adapt here to keep pairs with more / less levenshtein distance
  pull(row_num) # get the row_num variable as a vector

rows_to_keep
# we also want to get the word before each of the rows, so for each row number also keep the row before
# the c() means we "concatinate" the two vectors

rows_to_keep # we want this 
rows_to_keep - 1 # "plus" this
all_rows_to_keep <- c(rows_to_keep, rows_to_keep  - 1)

sort(all_rows_to_keep) # just to look at it

# get words!!
df %>% 
  filter(row_num %in% all_rows_to_keep) %>% 
  select(word)