I have worked years in Machine Learning and Natural Language Processing. I'm also familiar with Java and using SpringMVC for Restful API. For your problem, I did once by analyze sentiment in tweets from Twitter. The model I used in that project is Naive Bayes and it's really fast and acceptable accuracy. However, I think the accuracy can be improved by using more complex models, such as SVM or Random Forest.
Feel free to inbox me for further discussion.