UDPipe Natural Language Processing - Try it out

In udpipe: Tokenization, Parts of Speech Tagging, Lemmatization and Dependency Parsing with the 'UDPipe' 'NLP' Toolkit

options(width = 1000)
knitr::opts_chunk$set(echo = TRUE, message = FALSE, comment = NA, eval = TRUE)

Install the R package.

install.packages("udpipe")

Example

Get your language model and start annotating.

library(udpipe)
udmodel <- udpipe_download_model(language = "dutch")

knitr::opts_chunk$set(echo = TRUE, message = FALSE, comment = NA, eval = !udmodel$download_failed)

udmodel <- udpipe_load_model(file = udmodel$file_model)
x <- udpipe_annotate(udmodel, 
                     x = "Ik ging op reis en ik nam mee: mijn laptop, mijn zonnebril en goed humeur.")
x <- as.data.frame(x, detailed = TRUE)

Or just do as follows.

library(udpipe)
x <- udpipe(x = "Ik ging op reis en ik nam mee: mijn laptop, mijn zonnebril en goed humeur.",
            object = "dutch")

The annotation returns paragraphs, sentences, tokens, the location of the token in the original text, morphology elements like the lemma, the universal part of speech tag and the treebank-specific parts of speech tag, morphosyntactic features and returns as well the dependency relationship. More information at https://universaldependencies.org/guidelines.html

str(x)

A small note on encodings

Mark that it is important that the x argument to udpipe_annotate is in UTF-8 encoding. You can check the encoding of your text with Encoding('your text'). You can convert your text to UTF-8, using standard R utilities: as in iconv('your text', from = 'latin1', to = 'UTF-8') where you replace the from part with whichever encoding you have your text in, possible your computers default as defined in localeToCharset(). So annotation would look something like this if your text is not already in UTF-8 encoding:

• udpipe_annotate(udmodel, x = iconv('your text', to = 'UTF-8')) if your text is in the encoding of the current locale of your computer.
• udpipe_annotate(udmodel, x = iconv('your text', from = 'latin1', to = 'UTF-8')) if your text is in latin1 encoding.
• udpipe_annotate(udmodel, x = iconv('your text', from = 'CP949', to = 'UTF-8')) if your text is in CP949 encoding.

invisible(if(file.exists(udmodel$file)) file.remove(udmodel$file))

udpipe documentation built on Jan. 6, 2023, 5:06 p.m.