Replace a list of words with one unique word in R

Question

I am working on a text analysis with R and have a dataset (text corpus) with various sentences about different fruits. For example: "apple", "banana" , "orange", "pear", etc.

Since it is not relevant for the analysis whether someone writes about "apples" or "bananas", I want to replace all different fruits with one specific word, for example "allfruits".

I thought about using regex but I am facing two issues;

1) I want to avoid separate code lines for each kind of fruit. Thus, is there a way to define a list or a vector that I can use so that the function replaces all words in that list (apple, bananas, pear, etc.) with one specific word "allfruits"?

2) I want to avoid that words that are NOT a fruit but contain the same string as a fruit (e.g. the word "appletini) get replaced by the function.

Example: If I have a sentence that says: "Apple is my favourite fruit, appletini is my favourite drink. I also like bananas!" I want following to happen: allfruits is my favourite fruit, appletini is my favourite drink. I also like allfruits!

I am not sure whether it is possible to write this with a gsub function. Thus, all help is much appreciated.

Thank you!

score 1 · Accepted Answer · answered Jun 04 '20 at 20:36

1

allfruits can be extended to contain any words to be replaced:

allfruits = c("apple", "banana" , "orange", "pear")
replacement = "allfruits"
text = "Apple is my favourite fruit, appletini is my favourite drink. I also like bananas!"

gsub(paste0("\\b(", paste0(allfruits, collapse="|"), ")[s]?\\b"), replacement, text, ignore.case = TRUE)

Returns

[1] "allfruits is my favourite fruit, appletini is my favourite drink. I also like allfruits!"

The regex:

\\b - wordboundary
(", paste0(allfruits, collapse="|"), ") - all fruits names separated by a | (or)
s? - optional letter 's'
\\b - wordboundary
ignore.case = TRUE - ignore case

answered Jun 04 '20 at 20:36

dario

6,415
2
12
26

Thank you so much! This really helped to solve the issue. I am new to the whole gsub and regex functions topic, so it is still a bit confusing, especially the part about the "wordboundaries". Do you have an example of how I would write the regex if I wanted to replace a specific string within a word - for example to replace "apple" within the word "appletini" and make "allfruitstini" out of it?. Same goes for an example if the word apple was in between two strings (like: "stringapplestring" becomes "stringallfruitsstring") – lole_emily Jun 05 '20 at 17:09
Removing the wordboundary part in the example above would do what you ask in the comment. In the descripion of the `pattern` argument in `?gsub` we find a link to R's regex documentation. There we find the different expressions we can use. You could then start with a teststring and try different patterns. – dario Jun 05 '20 at 17:36

score 0 · Answer 2 · answered Jun 04 '20 at 20:33

str <- "Apple is my favourite fruit, appletini is my favourite drink. I also like bananas!"
gsub("(\\bapples?\\b)|(\\bbananas?\\b)", "allfruits", str, ignore.case = T)

\\b means boundary, that is the end of a word (punctuation, space, nothing after...)
| means OR
() defines a group
s? means with a s if possible

Replace a list of words with one unique word in R

2 Answers2