CARA enables accurate prediction of the sentiment of consumer opinions in online reviews. This flexible solution can help you capture the voice of the customer and identify detractors and promoters.
CARA is a toolchain for training a sentiment predictor. It works by combining lexical and machine learning approaches and integrating them with usergenerated content processing tools from the ADAPT Centre. When tuned to and tested on laptop and restaurant reviews, CARA has recorded best-in-class performance in international shared task competitions. In addition to offering increased accuracy in sentiment analysis, CARA is easy to use and provides a convenient entry point for building solutions. Built-in tooling also makes it full adaptable to new domains.
