Text-based communication and linguistic traces can reveal information about people’s thoughts, motives, and relationships, and predict their emotional state and interpersonal and group dynamics. This is useful in assessing current belief states and intent, and for forecasting future behaviors; however, developing tools to do this remains a significant challenge. To address this, IAI has been awarded a new contract entitled, “A Text Mining System for Modeling Social Dynamics in Groups.” A novel Group Forecast Analysis (GFA) framework will be developed for extracting metrics of group dynamics from discourses, and producing predictive models of social processes among individuals and organizations. The framework has three layers. First, the data collection and preparation layer will collect textual communication data for a given group, using in-situ socio-cultural data from unstructured information like social media statuses and online speeches. It will access the dynamic warehouse system with large-scale, real-time social media and online data, developed in IAI’s Human Social Cultural Behavior (HSCB) data collection project. Then the linguistic and discourse analysis layer will identify and extract linguistic features related to group intent and action forecasting using social language processing, and analyze them at the individual and group level. Finally the group forecast layer will utilize a prediction engine to correlate individual and group level language metrics with different types of group dynamics, and also the strength of the link between intent and action. The prediction engine incorporates a set of learning algorithms from IAI’s Agent-Based data Miner (ABMiner) platform, which has more than 400 algorithms for both supervised and un-supervised learning, and can optimally identify underlying group dynamics using data mining techniques.