Reverse Engineering With Grammatical and Visual Suppor
Title: Reverse Engineering With Grammatical and Visual Suppor
Speaker: Professor Kang ZHANG（University of Texas at Dallas）
Time: 3:30pm,Friday,June 4th
Venue: Lecture Room, Level 3 Building #5, Institute of Software, CAS
Abstract: Discovering program behaviors and functionalities is the primary activity of reverse engineering. Existing program analysis approaches have used text mining algorithms to infer behavior patterns or formal models from program execution. When one tries to identify the hierarchical composition of a program behavior at different abstraction levels, textual descriptions are not informative and expressive enough. To address this, we present a semi-automatic graph grammar approach to retrieving the hierarchical structure of the program behavior. The hierarchical structure is built on recurring substructures in a bottom-up fashion. We formulate the behavior discovery and verification problem as a graph grammar induction and parsing problem, i.e. automatically iteratively mining qualified patterns and then constructing graph rewriting rules. Furthermore, using the induced grammar to parse the behavioral structure of a new program could verify if the program has the same behavioral properties specified by the grammar. We also present a multi-plane visualization approach to support program behavioral analysis, realized in a tool called SoftLink. Experiments have been conducted on an open source project.
张康 博士 现任美国德克萨斯大学达拉斯分校 计算机科学系 终生正教授,可视计算实验室主任。同时还是该校计算机工程专业和地理信息系统专业的兼职教授。他于1982年2月在成都电讯工程学院(现改名为电子科技大 学)取得计算机工程学士学位,1990年在英国布莱顿大学取得计算机科学博士学位。现任Journal of Visual Languages and Computing编委及书评编辑, International Journal of Software Engineering and Knowledge Engineering编委, International Journal of Advanced Intelligence编委。发表过5本论著,40多篇期刊论文,和130多篇会议论文。张康教授的主要研究领域是可视化语言,信息可视化,计算机与美 学和艺术的关系, 软件工程,及Web。