Understanding the real relationships between words (rather than just statistically counting sentiment-carrying words) allows for creating meta sentiment language with a new set of operators, which solves ambiguity, various types of negation and often also contextual sentiment with unparalleled precision also giving the opportunity to analyze and aggregate answers to 'why?' question in NPS results.
Example:
Review: "She was great. Now she is boring and has become a poor actress."
Rule built using meta sentiment language:
{past tense}{good sentiment} + {present tense}{bad sentiment} = bad sentiment + sentiment change
If you would like to learn more about how Language Decoder can help with sentiment analysis and processing NPS results contact us and we will be happy to provide detailed information and access to live demo.