Book excerpt: Data containing geographical locations and time that associates with natural language texts, such as geotagged tweets, travel blogs, and crime reports are generally recognized as spatiotemporal semantic events. Many research fields have tried to gain valuable insights from these data and there have many techniques and methods are introduced in past decade. In computer science field, the study of spatiotemporal-semantic events in visualization and visual analytics is one of the hottest research topics. Text mining and data mining provide abundant methods to find meaningful knowledge and insights from semantic information of these data. Even though, there exist many contributions in this research field, there still lack of visually intuitive applications and approaches that allow frontline users, such as police, health officers, and social workers to freely navigate, effectively utilize and analyze their spatiotemporal semantic data, especially in community level. In this thesis, multiple visual analytics (VA) solutions are introduced. NeighborVis, CLEVis, and a new lens based visual interaction technique, GTMapLens to help frontline users harness semantic-rich spatiotemporal data. The development of all applications is fulfilled the requirement analysis and initial prototype evaluation. Text mining, topic modeling, hierarchical geospatial data indexing and many new visualization methods are studied and discussed along with those VA systems. The visual design is guided by requirement analysis with a cohort of multidisciplinary domain experts. Evaluation is presented with real world datasets to show the usability and effectiveness.