Enhanced Sequitur for finding structure in data
Eric A Earl, Richard E. Ladner · 2003
Summary form only given. The enhancements made to Sequitur for two more specific kinds of input are described, namely DNA strings and digitized music. Because of the natural orientation of the double helix DNA structure, DNA sequences bond to their reverse complements. Sequitur is enhanced to recognize reverse complements. These enhancements improved Sequitur's ability to compress and discover structure in DNA sequences. Several tables are given, which show a typical result for one human DNA sequence that is known to have some repetitive elements. Sequitur is enhanced to recognize three musical transformations: inversions, reversals, and transpositions.