Local sequence alignment as a method to detect temporal patterns in behavioral data

A time-series is a sequence of observations ordered by time. Often in the behavioral sciences, these observationsare instances of categorical variables and can be represented by a finite set of symbols, or an alphabet. In thesesequences, there may exist temporal patterns that are important in understanding the dynamics of behavior. However,these patterns may be nontrivial, that is events in the patterns may be noncontiguous and therefore difficult todetect by standard time-series analyses as these methods generally deal with understanding the structure of behaviorat a global level across the entirety of the series. In 1981, Temple Smith and Michael Waterman encountered asimilar issue in the field of molecular biology. They developed local sequence alignment as a means to discovernontrivial patterns of similarity in long sequences of DNA and protein, each comprised of elements from an alphabetof size four and twenty respectively. This project will describe methods of local sequence alignment as they existin the biological sciences and propose and implement analogous methods for use with temporal data.