Deep Learning A.I. that Extracts Meaning and Emotion from Text
After the first few weeks of running the Modulus Sentiment Analysis System back in 2009, we noticed something very peculiar.
The system began showing the overall mood of the world, by indicating the predominate emotion propagating through social media.
Love, joy, disgust, anger, sadness, fear, surprise... the system analyzed and recorded the world's emotions in real-time, second-by-second.
The Modulus Social Media Sentiment Analysis System uses to extract mood and emotions from millions of social media messages in real-time. After running uninterrupted for nearly a decade, the system has amassed multiple petabytes of valuable historic time-series data.
The system uses the Modulus Real Time Data Server for data storage.
In fact, Granger causality testing between our data and the DJIA have shown extremely high correlations. We have no doubt about professor Johan Bollen's claim that the system is capable of predicting the stock market up to three days in advance with an 87.6% accuracy rate.
"Twitter mood predicts the stock market". Journal of Computational Science 2: 1-8. Johan Bollen; Huina Mao; Xiao-Jun Zeng (2010).
Coincidentally, we developed our system one year prior to the aforementioned publication.
Some of the world's top performing hedge funds, as well as governmental organizations, have used our system for forecasting and analysis.
Our system uses deep learning neural networks with the help of OpenCyc and WordNet to extract the true meaning, intention, and emotion behind social media messages in ten languages, including English, Chinese (Mandarin), Spanish, Arabic, Hindi, Bengali, Portuguese, Russian, Japanese and German.
Our multiple petabytes of historic and real-time data contain time-series for global mood such as love, joy, surprise, anger, sadness, fear, and general mood. The emotion definitions are based partially on the work of Professor W. Gerrod Parrott.
"Emotions in Social Psychology". Philadelphia: Psychology Press. Parrott, W. G. (Ed.). (2001).