R: Replace Abbreviations\ Words

I have tried to resolve this problem all day but without any improvement.

I am trying to replace the following abbreviations into the following desired words in my dataset:

-Abbreviations: USA, H2O, Type 3, T3, bp

Desired words United States of America, Water, Type 3 Disease, Type 3 Disease, blood pressure

The input data is for example

[1] I have type 3, its considered the highest severe stage of the disease.
[2] Drinking more H2O will make your skin glow.
[3] Do I have T2 or T3? Please someone help.
[4] We don't have this on the USA but I've heard that will be available in the next 3 years.
[5] Having a high bp means that I will have to look after my diet?

The desired output is

[1] i have type 3 disease, its considered the highest severe stage of the disease.
[2] drinking more water will make your skin glow.
[3] do I have type 3 disease? please someone help.
[4] we don't have this in the united states of america but i've heard that will be available in the next 3 years.
[5] having a high blood pressure means that I will have to look after my diet?

I have tried the following code but without success:

   data= read.csv(C:"xxxxxxx, header= TRUE")
   lowercase= tolower(data$MESSAGE)
   dict=list("\\busa\\b"= "united states of america", "\\bh2o\\b"= 
   "water", "\\btype 3\\b|\\bt3\\"= "type 3 disease", "\\bbp\\b"= 
   "blood pressure")
   for(i in 1:length(dict1)){
   lowercasea= gsub(paste0("\\b", names(dict)[i], "\\b"), 
   dict[[i]], lowercase)}

I know that I am definitely doing something wrong. Could anyone guide me on this? Thank you in advance.

Solution

If you need to replace only whole words (e.g. bp in Some bp. and not in bpcatalogue) you will have to build a regular expression out of the abbreviations using word boundaries, and - since you have multiword abbreviations - also sort them by length in the descending order (or, e.g. type may trigger a replacement before type three).

An example code:

abbreviations <- c("USA", "H2O", "Type 3", "T3", "bp")
desired_words <- c("United States of America", "Water", "Type 3 Disease", "Type 3 Disease", "blood pressure")
df <- data.frame(abbreviations, desired_words, stringsAsFactors = FALSE)
x <- 'Abbreviations: USA, H2O, Type 3, T3, bp'
sort.by.length.desc <- function (v) v[order( -nchar(v)) ]

library(stringr)
str_replace_all(x, 
    paste0("\\b(",paste(sort.by.length.desc(abbreviations), collapse="|"), ")\\b"), 
    function(z) df$desired_words[df$abbreviations==z][[1]][1]
)

The paste0("\\b(",paste(sort.by.length.desc(abbreviations), collapse="|"), ")\\b") code creates a regex like \b(Type 3|USA|H2O|T3|bp)\b, it matches Type 3, or USA, etc. as whole word only as \b is a word boundary. If a match is found, stringr::str_replace_all replaces it with the corresponding desired_word.

See the R demo online.