SoD-HCER: Incorporating Uncertainty in the Evaluation of Software Designs

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

Directorate for Computer and Information Science and Engineering (CISE) Division Computer and Network Systems (CNS) Science of Design (SoD) Program Proposal Number: 0613823 P/I: Mary Shaw PI's Department: Computer Science Institution: Carnegie-Mellon University Award: $132,021 Title: "SoD-HCER: Incorporating Uncertainty in the Evaluation of Software Designs" This project involves the development of a systematic framework, the software confidence chain, for codifying and managing uncertainty in software design. This research lays the groundwork for confidence chains by designing and validating a representation for evaluation information that also captures confidence information and cost of analysis and that supports dynamic update of the information. In addition, the project will explore two specific techniques for handling uncertainty, characterizing uncertainty about requirements and resources that change over time and reducing uncertainty about end users' needs and expectations by improving ways for them to express their needs and expectations. The proposed research will provide a scientific basis for describing and managing uncertainty in software design, especially early in the design process, while it is still relatively inexpensive to make changes. Connecting assessments of uncertainty and confidence directly to analysis will improve the linkage to costs and benefits seen by clients, enabling more informed business decisions about software development. This project will improve our fundamental understanding of design by providing a unified approach to describing and managing uncertainty that recognizes and preserves the relations among sources of uncertainty and confidence. It will contribute analysis techniques that both provide useful analysis power and explore different facets of uncertainty, especially in the early stages of design. It will contribute both to a scientific understanding of specific types of uncertainty management and to the more general challenge of developing predictive evaluations. In addition, the research will contribute to a longer-term objective of developing a comprehensive model that rigorously relates a wide class of analyses. It will also help to establish an interdisciplinary bridge between software design and well-established business models. Development of a scientific basis for finding design points that are cost effective in light of uncertainty holds promise for improving products used by the community at large, for example by making it easier to select from among similar competing products. Similarly, increased attention to user preferences, especially for dynamic adaptation to preferences, should make computing less inscrutable to everyday users. Program Manager: Anita J. La Salle Date: July 5, 2006