CSC News
Xie Receives NSF Award to Improve Software System Reliability
Dr. Tao Xie has been awarded $20,000 by the National Science Foundation to fund his research proposal titled “CSR-SMA: Improving Software System Reliability via Mining Properties for Software Verification.”
The award will run from August 1, 2007 through July 31, 2008.
Research Abstract - Most correctness, security, and robustness violations of software systems are caused by the incorrect usage of application-specific APIs. But API details and the implicit usage properties are often not documented by the developers. Indeed, domain-specific or application-specific API properties or behaviors can be formally specified and statically verified against software systems. However, two major hindrances exist in using the current state of the art static verification tools. Manually specifying a large number of properties or behaviors for static verification is often inaccurate or incomplete, apart from being cumbersome and prohibitively expensive. In this project, we propose to develop a set of practical techniques and tools for inferring properties centered around single API call and properties related to multiple API calls, and improving the inference results through automatic test generation and dynamic analysis.
The award will run from August 1, 2007 through July 31, 2008.
Research Abstract - Most correctness, security, and robustness violations of software systems are caused by the incorrect usage of application-specific APIs. But API details and the implicit usage properties are often not documented by the developers. Indeed, domain-specific or application-specific API properties or behaviors can be formally specified and statically verified against software systems. However, two major hindrances exist in using the current state of the art static verification tools. Manually specifying a large number of properties or behaviors for static verification is often inaccurate or incomplete, apart from being cumbersome and prohibitively expensive. In this project, we propose to develop a set of practical techniques and tools for inferring properties centered around single API call and properties related to multiple API calls, and improving the inference results through automatic test generation and dynamic analysis.
Return To News Homepage