Sentiment Analysis using Stanford CoreNLP Recursive Deep Learning Models Sentiment analysis is usually carried out by defining a sentiment dictionary , tokenizing the text , arriving at scores for individual tokens and aggregating them to arrive at a final sentiment score. The limitation is that "not great" could be classified as neutral though it is clearly negative. Stanford's sentiment model uses phrases to identify the sentiments instead of words . It then builds a sentiment tree to compute the overall sentiment. More information can be found at http://nlp.stanford.
Read full article from Sentiment Analysis using Stanford CoreNLP Recursive Deep Learning Models | As I Know
No comments:
Post a Comment