A study of methods for textual satisfaction assessment

Software projects requiring satisfaction assessment are often large scale systems containing hundreds of requirements and design elements. These projects may exist within a high assurance domain where human lives and millions of dollars are at stake. Satisfaction assessment can help identify unsatisfied requirements early in the software development lifecycle, when issues can be corrected with less impact and lower cost. Manual satisfaction assessment is expensive both in terms of human effort and project cost. Automated satisfaction assessment assists requirements analysts during the satisfaction assessment process to more quickly determine satisfied requirements and to reduce the satisfaction assessment search space. This paper introduces two new automated satisfaction assessment techniques and empirically demonstrates their effectiveness, as well as validates two previously existing automated satisfaction assessment techniques. Validation shows that automatically generated satisfaction assessments have high accuracy, thus reducing the workload of the analyst in the satisfaction assessment process.