Identifying Software Theft Based on Classification of Multi-Attribute Features

Abstract

Due to the low performance caused by the traditional embedded watermark and the shortages about low accuracy and weak anti-aggressive of single-attribute birthmark in checking obfuscated software theft, a software identification scheme is proposed which is based on classification of multi-dimensional features. After disassembly analysis and static analysis on protecting software and its resisting semantics-preserving transformations, the algorithm extracts features from many dimensions, which combines the statistic and semantic features to reflect the behavior characteristic of the software, analyzing and detecting theft based on similarities of software instead of traditional ways depending on a trusted third party or alone-similarity threshold. Through giving the formal description about the algorithm, depicting the algorithm realization, after comparisons and analysis from the qualitative and quantitative, theoretical and experimental aspects, the results show that the algorithm contributes to the resistance to attacks, as well as the robustness and credibility, and has advantages compared with similar methods.