STM Article Repository

Successfully understanding of social media conversation growth, dissemination and extinction is a
challenging task that relies on identifying groups, group influence, diffusion models, forecast models, social
dynamics and text analytics. In this problem, we concentrate on the description of a novel approach for
identifying drivers of the direction and momentum of social conversations, including the spread of mood,
sentiment and issues. The approach first groups potential drivers of conversation based on variability. The
primary driver in each group is then selected. Finally, the relationship between the selected drivers and the topic
outcome is calculated and displayed visually. This enables the quick identification of the form and structure of
the conversation and allows us to predict momentum, direction, contagion risks, potential responses and
interventions. There is a huge amount of data in the form of text available today in the internet across various
channels – social media, news articles, blogs, e-commerce websites. Most of this data is a part of some
“conversation” or the other where real-world entities discuss, analyze, comment, exchange information in the
form of written expressions in textual format. Driver Modeling on textual data can be useful in observing the
key drivers which are driving the “conversation” coupled with the associated sentiments and mood states for the
observed key drivers. These insights about the key conversational drivers are often used in a variety of domains
such as tracking news cycles, stock movements, legislation developments, brand image, viral breakouts and
much more.