StarSPIRE is a visual analytics tool prototype specializing in unstructured text analysis. StarSPIRE [Figure 1] is a multi-model semantic interaction tool, which couples foraging and synthesis activities through learning from user interactions to adjust a model of the user's interests within the dataset. This model is then used to rearrange data, represented as a node-link diagram, in the workspace as well as determine what data should be displayed and what should be stored out of sight. This push-and-pull nature of disclosing only a portion of the dataset that exceeds a variable relevance threshold allows the user to explore the dataset incrementally and according to their current interests. Visual encodings such a node, size, color, and saturation are used to denote the perceived relevance of the document in order to guide the user through the documents. StarSPIRE is best designed for directed sensemaking tasks, as it presents the user with a blank screen and relies on a search to gather initial data onto the workspace.