Well-run organizations collect, archive and analyze data relating to the effectiveness of their important processes. Educational institutions discard a wealth of student scores that could be analyzed. Each score contains important information about the student as well as the item (i.e., problem or question). This paper describes our project to develop an outcomes-based assessment system that mines per-item scores to track each student's skills and knowledge. Statistical inference techniques from both educational statistics and data mining will quantitatively determine each student's acquired competency, with minimal input from faculty. The culmination of item-level assessment gives individual faculty feedback on their courses, and gives curriculum committees feedback on which objectives are sufficiently met by their respective curricula.