Modeling of the optimal sequence alignment problem

The sequence alignmet is widely used in variouse computer systems for data similarity measure and analisys, changes detection and relative tasks. Some intuitive requirements for string alignmet are formalised in a test suite. The tests shows that none of existing approaches to string similarities and alingment meet the requirements. A new model of minimizing conflicts when merging changes is described. The model leads to a simple formulation of new optimization problem which meet the requirements. (In Russian.)