How can I search for the variation of a word without typing all variations ( in R)?

Question

I need to check whether the variation of a word is in the text? How can I do that without typing everything out? For example, I need to search for the word 'broken', is there a way in r where it can look for the word and other variations?

a="Broken flask"
b="fragmented flask"
c="broke glass"
d="shattered glass"
e="break flask"
text=c(a,b,c,d,e)
str_detect(tolower(text),"broken|fragmented|broke|break|shatter|shattered")

I mean the synonyms of the word broken, and the various ways you can say it. Also is there a way I don't have specify the past, present of a specific word? — Ashti, May 10 '22 at 20:37
No, regex is not an NLP tool. You cannot use it here. Unless you add all those variations as alternatives. — Wiktor Stribiżew, May 10 '22 at 20:38
So there is no easier way to do it unless I specify the words I'm looking for? — Ashti, May 10 '22 at 20:40
`adist` is a quick and dirty way to do it. Ensure that the input is `tolower` and trimmed. — Mossa, May 10 '22 at 21:17
I don't think `adist` will work - OP is looking for *synonyms*, not spelling variations/typographical errors ... — Ben Bolker, May 10 '22 at 21:40
I don't see why this was "closed for more focus"? OP has given a fine MCVE ... ? — Ben Bolker, May 10 '22 at 21:43
@BenBolker I was wondering the same. Perhaps it was initially tagged with regex and the vtcs were due to it not being a regex problem? — Allan Cameron, May 10 '22 at 21:55

score 1 · Accepted Answer · answered May 10 '22 at 21:31

1

You could check out syn from the syn package, which generates synonyms for a given word, allowing you to do:

library(syn)

grepl(paste0(c("broken", syn("broken")), collapse = "|"), text, ignore.case = T)
#> [1]  TRUE  TRUE  TRUE  TRUE FALSE

It picked up 4 out of 5 here, without having to program any variations.

answered May 10 '22 at 21:31

Allan Cameron

147,086
7
49
87

1

You *might* be able to do a little bit better with a **stemmer**, see `corpus::text_tokens()` – Ben Bolker May 10 '22 at 21:48

How can I search for the variation of a word without typing all variations ( in R)?

1 Answers1