Stanford NLP Sentiment Analysis with entity recognition

3
votes

I am new to the field of Sentiment Analysis and I would like your help. What I need to do with Stanford Core NLP is provide raw text and get 2 kinds of results:

  1. The total sentiment of the whole text (e.g. positive, neutral or negative)

  2. Entity-based sentiment (either call a function to return a list with all the entities and their sentiment, or call a function where I provide the needed entity as an argument and it returns the sentiment related to this entity).

1
stanford-nlpsentiment-analysisnamed-entity-recognition
On stackoverflow it is generally appreciated if you show what you have tried, instead of just asking for a complete sollution. – DB5
Well as I said I very new to this. I have tried importing Stanford Core NLP into eclipse to fiddle around with it, but I cannot even build it because the "build all" option is not active and I do not know why. All I have managed so far is to execute the pre-built SentimentPipeline module, which produces the sentiment for each sentence of the input. – Nick Pavlakis
I finally managed to import CoreNLP into Eclipse but I cannot figure out how to use it to perform sentiment analysis (apart from running the pre-built demo). Could any more experienced user provide some sample code to start or point me to the right direction? – Nick Pavlakis
Still need help with this? – krinker

1 Answers

1
votes

I ran into a similar issue and I was wrapping several Java projects in an easy to use webservice. I was using Google App Engine and you can find the code here: https://github.com/josepvalls/parserservices

I added the Stanford Parser, Berkeley Parser, the Stanford CoreNLP and the OpenNLP Parser, unfortunately I never got to NER.

On a related note, if you can use Stanford NER instead of OpenNLP NER, there is an online demo that you can use: http://nlp.stanford.edu:8080/ner/