Semi-structured documents and data pervade modern workflows in all areas. Collaborative work and version management rely on effective, automatic difference analysis and three-way difference analysis tools. In our effort to develop a three-way difference analysis for tree-structured documents we developed a kernel three-way difference algorithm which extends the equality-based procedures, such as GNU diff3, by considering the similarity of documents in the difference analysis as well as to ignore the order of data if that is semantically suitable. As a result we obtain difference analysis algorithms that can be more fine-tuned to the application domain. Moreover, the equality-based counter-parts of our three-way difference analysis algorithms has the idempotency property, which the current three-way diff algorithms lacks.
Serge Autexier. Similarity-Based Diff, Three-Way Diff and Merge. International Journal of Software and Informatics, 2015,9(2):259~277Copy