no issue, just congratulations!

[randomgambit]

amazing job, amazing package. I love the as.data.frame function (can you use dplyr::data_frame as well?). I hope the package will get better and better. what are the next updates?

Thanks!!

[dselivanov]

no, no, no dplyr please. @randomgambit you can use it on your own if you really like it.

[jwijffels]

I have not plans to make this a dplyr dependency, on the contrary. I don't like the tidyverse package dependencies and was already reluctant to add the data.table dependency.

Next on my todo list is:

• allowing only parts of the annotation process (e.g. only tokenising, tagging or dependency parsing)
• getting the package on cran
• making models available outside cran (as they are too big in size) in other R packages for convenience
• allowing to train models based on conll-u data directly
• try to remove the data.table dependency again without speed sacrifices

[randomgambit]

@jwijffels OK OK but please consider that whether you like it or not, many users use either dplyr or data.table. Something I really like is outputting the POS tagging into a data.frame for further use. Removing the data.table thing would be a big mistake in my opinion.

[dselivanov]

No one prevent of the usage of favorite tools. The thing is that data.frames are base R data structures - stable and tested. What @jwijffels have told - is that he doesn't want to have dependencies on these packages. He haven't told that he wants to remove as.data.frame. I also try to keep number of dependencies small to some extent.

[jwijffels]

@randomgambit You can just use tibble::as.tibble on the output of the as.data.frame function to get it in dplyr format if you need that for your later processing.

[randomgambit]

@dselivanov @jwijffels OK perfect then! :)