Identifying and Forecasting Shifts in the Mood of Social Media Users

Quantitatively identifying and forecasting shifts in a mood of social media users is described. An example method includes categorizing the textual messages generated from the social media users over a selected period of time into a plurality of word categories, with each word category containing a set of words associated with the mood of social media users. A score indicating an intensity of the mood of the social media users is calculated for each word category, wherein a value of the score and its corresponding time point define a data point for the word category. Subsequently, breakpoints in the mood of social media users are determined so that the breakpoints minimize a sum of square errors representing a measurement of a consistency of all data points from inferred values of the scores of the data points derived using the breakpoints over the selected period of time. Further, space of all possible breakpoints for the word categories are searched to identify a defined number and locations of the breakpoints. Finally the breakpoints over the selected period of time are interpreted to identify the shifts in the mood of social media users and trends between breakpoints.