ABSTRACT
In our discussion, we present initial findings from a research effort involving population sentiment analysis as part of the Social Network Analysis Reachback Capability (SNARC) project lead by MITRE in support of the Network Effects Cell at the Information Dominance Center (IDC) International Security Assistance Force (ISAF) Joint Command (JC). In support of SNARC, we performed a multidimensional analysis of attitudinal survey data from Afghanistan using an OLAP (online analytical processing) platform. We also conducted a Bayesian analysis of the data to determine influence patterns within the data and identify significant trends relevant for operations. As part of this effort, we developed a Sentiment Prediction Model that predicts responses for locations where it is not feasible or desirable to directly interview the population of interest. Our hypothesis is that as part of this study, we will be able to identify the most relevant attributes that can help determine responses for a set of survey questions, with a certain degree of confidence. Our results show the utility in multidimensional analyses and influence modeling of survey data in response to requests for information.
