CODEWITHSUNDEEP
rtext-mining
Reuben Kandiah
Reuben Kandiah

Reputation: 51

R: Replace Substrings in Data Frame using a Lookup Table

I have a dataframe containing a series of strings, some of which contain two-word phrases which I want to condense down to a single "pseudo word".

In this example, "united kingdom","saudi arabia", and "european union" are phrases which are of interest to me. I would like to replace all instances of "united kingdom" with "unitedkingdom", "saudi arabia" with "saudiarabia" and so on.

My dataframe of text strings is as follows:

text.df <- as.data.frame(
c(
  "Lorem ipsum dolor sit saudi arabia adipiscing elit.", # Contains "saudi arabia"
  "Ut enim ad minim veniam united kingdom exercitation.", # Contains "united kingdom"
  "Excepteur sint european union deserunt saudi arabia laborum", # Contains "european union" and "saudi arabia"
  "Sed ut perspiciatis unde omnis error sit voluptate." # Contains nothing of interest
 ))
colnames(text.df) <- 'content'

My lookup dataframe is as follows:

lookup <- data.frame(matrix(ncol = 2, nrow = 3))
lookup$X1 <- c('united kingdom', 'european union', 'saudi arabia')
lookup$X2 <- c('unitedkingdom', 'europeanunion', 'saudiarabia')

My aim is to return a dataframe which looks like:

> new.text.df
                                                    content
1        Lorem ipsum dolor sit saudiarabia adipiscing elit.
2       Ut enim ad minim veniam unitedkingdom exercitation.
3 Excepteur sint europeanunion deserunt saudiarabia laborum
4       Sed ut perspiciatis unde omnis error sit voluptate.
>

If anyone is able to help it would be greatly appreciated! Thanks in advance.

Upvotes: 2

Views: 73

Answers (2)

Bechir Barkallah
Bechir Barkallah

Reputation: 341

You could try this :

library(stringr)
transform_word <- function(text){
 for (i in 1:nrow(lookup))
  text <- stringr::str_replace_all(text,lookup$X1[i],lookup$X2[i])
 return(text)
}

text.df[,'content'] <- sapply(text.df[,'content'],transform_word)

Upvotes: 1

AntoniosK
AntoniosK

Reputation: 16121

library(qdap)

text.df <- as.data.frame(
  c(
    "Lorem ipsum dolor sit saudi arabia adipiscing elit.", # Contains "saudi arabia"
    "Ut enim ad minim veniam united kingdom exercitation.", # Contains "united kingdom"
    "Excepteur sint european union deserunt saudi arabia laborum", # Contains "european union" and "saudi arabia"
    "Sed ut perspiciatis unde omnis error sit voluptate." # Contains nothing of interest
  ), stringsAsFactors = F)
colnames(text.df) <- 'content'

lookup <- data.frame(matrix(ncol = 2, nrow = 3))
lookup$X1 <- c('united kingdom', 'european union', 'saudi arabia')
lookup$X2 <- c('unitedkingdom', 'europeanunion', 'saudiarabia')

# provide patterns, replacements, actual texts to update
mgsub(lookup$X1, lookup$X2, text.df$content)

# [1] "Lorem ipsum dolor sit saudiarabia adipiscing elit."       
# [2] "Ut enim ad minim veniam unitedkingdom exercitation."      
# [3] "Excepteur sint europeanunion deserunt saudiarabia laborum"
# [4] "Sed ut perspiciatis unde omnis error sit voluptate."

Upvotes: 3

Related Questions