The Wealth Management business had a big joint venture and business was spending an enormous amount of time and money to understand their financial advisor opinions about the new platforms, products and services.Sentiment Analysis had to beused to determine sentiments, emotions and attitudes of the FA. The text used for analysis can range from big document (e.g. Platformreviews,support tickets) to small status message (e.g. collaborating platform status messages) to determine:
- User behaviour
- Product feedback
- User intentions
Tungsten's sentiment engine extracts and analysed sentiments for a given product and feature set. Lexicon was created using:
- Common or default sentiment words
- Negation Words
- Blind Negation Words
- Split words
On this lexicon, series of pre-processing steps were performed namely Pos tagging, stemming, exaggerated word shortening, emoticon detections, hash tag processing.
And then Sentiment calculation was done using various text analytics algorithms to generate clustered sentiments across various operating modules of platforms and products.