Data available in social media, such as Facebook, and Twitter provide a wealth of unstructured data about many topics and interests, with one of the most widely followed being that of public, political figures. This data is important for a number of applications. For example, diplomats desire a good understanding of the people with whom they must negotiate. Collating and summarizing relevant, unstructured information is time-intensive, making it useful to develop methods to semi-automatically mine such information. Institution theory, specifically the concept of isomorphism, predicts that public figures adopt specific ritualistic styles of communication, which cross international boundaries. Therefore, it should be possible to develop rules and heuristics that can be applied consistently across nations to semi-automatically text mine social media postings of public figures to extract useful information. This research develops and applies heuristics for text mining vast amounts of data on public figures from social media. To demonstrate the feasibility of doing so, the Facebook pages of five Foreign Ministers (or equivalents) are mined to make inferences about where they are, and with whom they interact on specific days. An assessment is made of how generalizable the results are and how they contribute to our understanding of domain-specific text mining on large amounts of data.