CODEWITHSUNDEEP
stanford-nlp
rst.jobs
rst.jobs

Reputation: 21

Stanford Core NLP Sentiment Analysis: Training with my own data

I am working on sentiment analysis using coreNLP, I have some questions about training with my own dataset, it would be a great help if someone was able to give me some idea.

According to https://nlp.stanford.edu/sentiment/code.html to train ones own dataset

java -mx8g edu.stanford.nlp.sentiment.SentimentTraining -numHid 25 -trainPath train.txt -devPath dev.txt -train -model model.ser.gz

What's the dev.txt and what data do I need to add in this file? Also I have checked class PTBTokenizer but I didn't find any text2PTB token so I can train my data?

Can someone tell me how can I train with my data?

for example test data

Upvotes: 1

Views: 564

Answers (1)

rst.jobs
rst.jobs

Reputation: 21

I have found the answer which works for me call 

java -cp "*" -mx5g edu.stanford.nlp.sentiment.BuildBinarizedDataset -input sample.txt

sample.text would contain training data, Example 1 Today is not a good day. 3 good 3 good day 3 a good day this will generate

(1 (1 Today) (1 (1 (1 (1 is) (1 not)) (3 (1 a) (3 (3 good) (1 day)))) (1 .)))

Upvotes: 1

Related Questions