2

I am learning sentiment analysis in R using tidytext package. However, i am unable to set nrc as lexicon. Whenever i type get_sentiments ("nrc"), the above error is displayed. It says that lexicon coud only be "afinn", "bing" or "loughran". I tried updating the package (tidytext) and also installed 'syuzhet' package but still the problem exists. Please help!

AhmadAli
  • 21
  • 1
  • 1
    Please make your question reproducible by adding sample code and data that can **reproduce the problem**. – NelsonGon Jul 05 '19 at 12:27
  • There are only three lexicons as the error states: `"afinn", "bing" or "loughran".` which can be seen in the arguments although I do recall using `nrc` some time. Perhaps it was replaced. No mention of it in the docs: https://cran.r-project.org/web/packages/tidytext/tidytext.pdf – NelsonGon Jul 05 '19 at 12:28
  • It was removed. See [here](https://github.com/juliasilge/tidytext/search?q=nrc&unscoped_q=nrc). If you really need to use it, just use an older version of the package. – NelsonGon Jul 05 '19 at 12:35
  • I simply loaded tidytext package and accessed the 'sentiments' dataset from the library. According to the book, https://www.tidytextmining.com/sentiment.html , the sentiments dataset should be a 27,314 x 4 tibble. However, when i load it, R shows it to be a 6,786 x 2 tibble. – AhmadAli Jul 05 '19 at 17:59
  • I read the CRAN page on tidytext and it also says that lexicon could only be "affin", "bing" or "loughran". I am using tidytext version 0.2.1 – AhmadAli Jul 05 '19 at 18:02
  • The book is behind the latest in the package. NRC was removed for licensing reasons. Sentiments only has the `bing` lexicon. – NelsonGon Jul 05 '19 at 18:02
  • Alright! Thanks for your help. :) – AhmadAli Jul 05 '19 at 18:04
  • Cool then,nice day. – NelsonGon Jul 05 '19 at 18:08

1 Answers1

0

You can just replace;

nrc_joy <- get_sentiments("nrc") %>% 
  filter(sentiment == "joy")

with;

nrc_joy <- get_sentiments("afinn") %>% 
  filter(value == 2 |value == 3 | value == 4 | value == 5)
IHADDADEN MOHAMED EL FODIL
  • 1
  • 2