4
UNSTRUCTURED DATA: HOW TO TAME THE BEAST?
EMAlpha sentiment technology
Sentiment Classifier: We use a standard sentiment classifier. The classifier can use both static dictionaries supplied by the user, as well as create custom dictionaries. The underlying method is similar to what you might read in a standard textbook where the sentiment classifier is typically based on support vector machines. The sentiment classifier can automatically develop a sentiment dictionary as it gets feedback from the end-user. Our dictionary starting point is a simple one based on Loughran-Mcdonald (https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1331573) and our own India-specific proprietary dictionary developed by the EMAlpha Research Team. As we cover ...
