The likes, comments and posts we share on social media can often seem inconsequential, but they matter. They tap into some of the very elements that make us human, our addictions, desires, anxieties and joys.
What if we could understand the psychology of social media and use that knowledge to enrich other user interacting apps and products eg more human chat bot, public opinion detection about political tendencies, brand management ,stock market monitoring etc. This explain the importance of emotion detection extraction from these social platforms.
Emotion detection is typically modeled as a supervised multi -label classification problem, also because each post may express multiple emotions. Eulerslab is working on a paper where we implement a deep learning model to improve upon existing models of multi-label emotion classification which aims to extract all emotional context that Plutchik[fig below] mention in his psycho evolutionary theory of basic emotions (joy, sadness, anger, fear, trust, disgust, surprise and anticipation) plus optimism, pessimism and love.
We believe this work can be transformed into lot of use cases for Product/Service based companies for eg analyzing reviews of any app across play store wherein deeper and meaningful conclusions can be derived then just start rating given out by play store. Another use case is to keep tab on user feedback across social media . For Marketing guys, analyzing brand reputation and performing competitors analysis are direct use case of the solution .
Our analyses aren’t limited to reviews and social media; the EulersLab solution applies NLP for topic, keyword, and sentiment extraction from survey responses ,user emails and call-center text
Our work goes deeper than the traditional classification of positive,negative or neutral and as stated above classified all possible emotions thereby providing great insight.
Whereas our work focus currently on Sentiment Analysis, however without applying Content Analysis to it , we realize such a model will have little utility. However here comes difficult part, context analysis has to vary with domain for eg when it comes to a App , phrases like app is slow ,battery issue,uninstall etc are of importance to all developers .Building contextual awareness and coupling it with Sentimental Analysis is one of the key attribute of model developed by Eulerslab.