In this memoir , I examine different aspects of a simple but ubiquitous computer object: the string or sequence of symbols. The string of characters concept is at the crossroads of areas as information theory and language theory. Although simple, this notion is fundamental: data will always be represented, encoded and stored in computers as sequences of symbols at one time or another The increasing amount of information and data to which we have access, such as genomes of individuals or scanned documents, justifies that the algorithms and data structures handling them need to be optimized. Consequently, rigourous analysis is needed in order to guide the end-user and the designer of programs that manipulate these data. The average analysis is particularly appropriate here because the data reach such variety and large volumes that the typical case is best reflected than with the more usual worst case complexity. This obviously raises the very difficult problem of data modeling. Indeed in our setting we want two contradictory things: a model closer to the data, which really translates their specificities, but also a model yielding results, that is to say, able to predict the performance. Methods are most often those of analytic combinatorics and use a mathematical object , generating functions, to carry out the analyzes.
