### Abstract

We initiate a new class of string matching problems called Substring Compression Problems. Given a string S that may be preprocessed, the problem is to quickly find the compressed representation or the compressed size of any query substring of 5 (Substring Compression Query or SCQ) or to find the length i substring of S whose compression is the least (Least Compressible Substring or LCS problem). Starting from the seminal paper of Lempel and Ziv over 25 years ago, many different methods have emerged for compressing entire strings. Determining substring compressibility is a natural variant that is combinatorially and algorithmically challenging, yet surprisingly has not been studied before. In addition, compressibility of strings is emerging as a tool to compare biological sequences and analyze their information content. However, typically, the compressibility of the entire sequence is not as informative as that of portions of the sequences. Thus substring compressibility may be a more suitable basis for sequence analysis. We present the first known, nearly optimal algorithms for substring compression problems - SCQ, LCS and their generalizations - that are exact or provably approximate. Our exact algorithms exploit the structure in strings via suffix trees and our approximate algorithms rely on new relationships we find between Lempel-Ziv compression and string parsings.

Original language | English (US) |
---|---|

Pages | 321-330 |

Number of pages | 10 |

State | Published - Jul 1 2005 |

Event | Sixteenth Annual ACM-SIAM Symposium on Discrete Algorithms - Vancouver, BC, United States Duration: Jan 23 2005 → Jan 25 2005 |

### Other

Other | Sixteenth Annual ACM-SIAM Symposium on Discrete Algorithms |
---|---|

Country | United States |

City | Vancouver, BC |

Period | 1/23/05 → 1/25/05 |

### Fingerprint

### ASJC Scopus subject areas

- Software
- Mathematics(all)

### Cite this

*Substring compression problems*. 321-330. Paper presented at Sixteenth Annual ACM-SIAM Symposium on Discrete Algorithms, Vancouver, BC, United States.

**Substring compression problems.** / Cormode, Graham; Muthukrishnan, Shanmugavelayutham.

Research output: Contribution to conference › Paper

}

TY - CONF

T1 - Substring compression problems

AU - Cormode, Graham

AU - Muthukrishnan, Shanmugavelayutham

PY - 2005/7/1

Y1 - 2005/7/1

N2 - We initiate a new class of string matching problems called Substring Compression Problems. Given a string S that may be preprocessed, the problem is to quickly find the compressed representation or the compressed size of any query substring of 5 (Substring Compression Query or SCQ) or to find the length i substring of S whose compression is the least (Least Compressible Substring or LCS problem). Starting from the seminal paper of Lempel and Ziv over 25 years ago, many different methods have emerged for compressing entire strings. Determining substring compressibility is a natural variant that is combinatorially and algorithmically challenging, yet surprisingly has not been studied before. In addition, compressibility of strings is emerging as a tool to compare biological sequences and analyze their information content. However, typically, the compressibility of the entire sequence is not as informative as that of portions of the sequences. Thus substring compressibility may be a more suitable basis for sequence analysis. We present the first known, nearly optimal algorithms for substring compression problems - SCQ, LCS and their generalizations - that are exact or provably approximate. Our exact algorithms exploit the structure in strings via suffix trees and our approximate algorithms rely on new relationships we find between Lempel-Ziv compression and string parsings.

AB - We initiate a new class of string matching problems called Substring Compression Problems. Given a string S that may be preprocessed, the problem is to quickly find the compressed representation or the compressed size of any query substring of 5 (Substring Compression Query or SCQ) or to find the length i substring of S whose compression is the least (Least Compressible Substring or LCS problem). Starting from the seminal paper of Lempel and Ziv over 25 years ago, many different methods have emerged for compressing entire strings. Determining substring compressibility is a natural variant that is combinatorially and algorithmically challenging, yet surprisingly has not been studied before. In addition, compressibility of strings is emerging as a tool to compare biological sequences and analyze their information content. However, typically, the compressibility of the entire sequence is not as informative as that of portions of the sequences. Thus substring compressibility may be a more suitable basis for sequence analysis. We present the first known, nearly optimal algorithms for substring compression problems - SCQ, LCS and their generalizations - that are exact or provably approximate. Our exact algorithms exploit the structure in strings via suffix trees and our approximate algorithms rely on new relationships we find between Lempel-Ziv compression and string parsings.

UR - http://www.scopus.com/inward/record.url?scp=20744439529&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=20744439529&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:20744439529

SP - 321

EP - 330

ER -