The q-gram filter is a popular filtering method for approximate string matching. It compares substrings of length q (the q-grams) in the pattern and the text to identify the text areas that might contain a match. A generalization of the method is to use gapped q-grams, subsets of q characters in some fixed non-contiguous shape, instead of contiguous substrings. Although mentioned a few times in the literature, this generalization has never been studied in any depth. In this paper, we report the first results from a study on gapped q-grams. We show that gapped q-grams can provide orders of magnitude faster and/or more efficient filtering than contiguous q-grams. The performance, however, depends on the shape of the q-grams. The best shapes are rare and often possess no apparent regularity. We show how to recognize good shapes and demonstrate with experiments their advantage over both contiguous and average shapes. We concentrate here on the k mismatches problem, but also outline an approach for extending the results to the more common k differences problem.
CITATION STYLE
Burkhardt, S., & Kärkkäinen, J. (2001). Better filtering with gapped q-grams. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 2089, pp. 73–85). Springer Verlag. https://doi.org/10.1007/3-540-48194-x_6
Mendeley helps you to discover research relevant for your work.