[stansent in Parallel?]

Open iamkbpark opened this issue 9 years ago • 1 comments

Dear Mr. Rinker:

First of all, I would like to express my sincere appreciation for 'stansent' as well as 'sentimentr'. Your packages makes me way easier to do sentiment analysis. Again, I do appreciate that.

Now, I just wonder if there is any way to use stansent with Stanford CoreNLP in parallel. So far, I have successfully run the analysis with my laptop as follows:

tagger_path = 'INSTALLED LOCATION sentiment_stanford_by (REVIEW, stanford.tagger = tagger_path)

I am currently testing the sample size of 10,000 reviews to get sentiment score but soon plan to increase the size of 500,000 reviews in AWS.

Thank you for your time.

Dec 18 '16 03:12 iamkbpark

This sounds like it would be useful. There's nothing built in. Perhaps you could break the task up into chunks and use R's parallel package? You might run into java memory issues. If that works for you and you let me know it could be a nice feature to include out of the box.

Dec 18 '16 22:12 trinker