Software Therapist: Usability Problem Diagnosis Through Latent Semantic Analysis

摘要

The work we report on here addresses the problem of low return on investment in software usability engineering and offers support for usability practitioners in identifying, understanding, documenting, and fixing usability problems. It does this by (1) validating and extending a structured knowledge framework of usability concepts for organizing and relating usability data to design flaws and solutions, (2) specifying a usability data management cycle to support a diagnosis process that is iteratively interleaved with data collection, and (3) developing a software system for usability engineering practitioners that includes components to support the activities of usability data collection, organization, and problem analysis and reporting, as well as automated support for usability problem diagnosis using a sophisticated statistical technique for the analysis of text. In this report, we briefly review the motivations and background for this work and describe the User Action Framework (UAF) and Latent Semantic Analysis (LSA); describe the design, development, and use of the Software Therapist system; discuss the research done on applying LSA to usability engineering; describe the usability content library collected as part of the project; and discuss the evaluation of the system as well as some lessons learned and possible future directions resulting from the project.