search

statsmaths / cleanNLP

R package providing annotators and a normalized data model for natural language processing
GNU Lesser General Public License v2.1
209 stars 36 forks source link

Error thrown when installing cleannlp #69

Closed tshjortile closed 4 years ago

tshjortile commented 4 years ago

Hi! I hope my issue is in the right place here, but since I did not find any solution to this issue so far I was wondering if you could help: I'm doing a digi hum project using cleannlp with spacy in R, and in my project setup I have run into some issues when installing cleannlp via pip install, which fails to complete due to the following errors:

ERROR: Could not find a version that satisfies the requirement torch>=1.0.0 (from stanfordnlp->cleannlp) (from versions: 0.1.2, 0.1.2.post1, 0.1.2.post2) ERROR: No matching distribution found for torch>=1.0.0 (from stanfordnlp->cleannlp)

I installed the latest version of torch (1.3.1) and have python 3.7.

Thank you! E.

statsmaths commented 4 years ago

Sorry, just saw this but it was closed. I assume that the issue was resolved? The current problem is that the python package requires both spacy and CoreNLP, even if you only want one (usually not an issue, but I think CoreNLP has some trouble on Windows machines). I need to fix this in the upcoming release.

tshjortile commented 4 years ago

Hi! I decided on a different approach as I was not able to solve the problem I described: In the end it lied with the coreNLP, which was too difficult to resolve, at least for me. Thank you for getting back to my request though, I hope to be able to using cleanNLP in the future!

Best wishes

On Wed, Jan 8, 2020, 12:11 PM Taylor Arnold notifications@github.com wrote:

Sorry, just saw this but it was closed. I assume that the issue was resolved? The current problem is that the python package requires both spacy and CoreNLP, even if you only want one (usually not an issue, but I think CoreNLP has some trouble on Windows machines).

— You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub https://github.com/statsmaths/cleanNLP/issues/69?email_source=notifications&email_token=AI6YEIKRZE4A3P4LQOLEL33Q4WYF7A5CNFSM4KC3DCZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIMBGHQ#issuecomment-572003102, or unsubscribe https://github.com/notifications/unsubscribe-auth/AI6YEIOPEO62O6NSQXGOKCLQ4WYF7ANCNFSM4KC3DCZA .

statsmaths commented 4 years ago

Ah okay, great though sorry that it was such an issue. Can I ask why you were trying to use CoreNLP rather than one of the other backends? I get a lot of questions and issues with CoreNLP, but I am not sure why its so popular in comparison to the UDPipe (super easy to install, lots of langs) and Spacy (more accurate the CoreNLP, easier to install, fastest implementation).

tshjortile commented 4 years ago

As you probably guessed I am pretty new to text analysis / topic modelling, and I only mastered R so far in regards to programming languages, so I was looking for digital humanities tutorials that would give me an overview how I could approach topic modelling in R using NLP. I found this one, which includes your CleanNLP - which is the reason I wanted to try it as well: https://susie-kim.github.io/post/2018-01-09-guide-cnlp-part1/ I realised after a while that I don't need NLP for what I aim to do, and since I had some problems installing CoreNLP on my windows machine, I looked for simpler solutions. I hope this helps in any way!