Increasing the usability of software by integrating usability engineering into the development cycle has become common practice. Although usability engineering is effective, it can be expensive, and organizations want to receive the best possible returns on their investments. Oftentimes, however, organizations spend large sums of money collecting usability problem data through activities such as usability testing, but do not receive acceptable returns on those investments during redesign. The primary reason is that there is an almost complete lack of methods and tools for usability problem data analysis to transform raw usability data into effective inputs for developers. In this thesis, we develop an infrastructure for usability problem data analysis to address the need for better returns on usability engineering investments. The infrastructure consists of four main components: a framework, a process, tools, and semantic analysis technology. Embedded within the infrastructure is the User Action Framework, a conceptual framework of usability concepts, which is used to organize usability data. The process addresses extraction of usability problems from raw usability data, diagnosis of problems according to usability concepts, and reporting of problems in a form that is usable by developers. The tools leverage the framework and guide practitioners through the process, while the semantic analysis technology supplements the capabilities of the tools to automate parts of the process.