Sequence alignments are used to capture patterns composed of elements representing multiple conceptual levels through the alignment of sequences that contain overlapping and variable length annotations. The alignments also determine the proper context window of words and phrases that most directly impact the meaning of a given target within a sentence, eliminating the need to predefine a fixed context window of words surrounding the targets. We evaluated the system using the CoNLL-2003 named entity recognition (NER) task.