### Abstract

Emerging data stream management systems approach the challenge of massive data distributions which arrive at high speeds while there is only small storage by summarizing and mining the distributions using samples or sketches. However, data distributions can be "viewed" in different ways. A data stream of integer values can be viewed either as the forward distribution f(x), ie., the number of occurrences of x in the stream, or as its inverse, f ^{-1}(i), which is the number of items that appear i times. While both such "views" are equivalent in stored data systems, over data streams that entail approximations, they may be significantly different. In other words, samples and sketches developed for the forward distribution may be ineffective for summarizing or mining the inverse distribution. Yet, many applications such as IP traffic monitoring naturally rely on mining inverse distributions. We formalize the problems of managing and mining inverse distributions and show provable differences between summarizing the forward distribution vs the inverse distribution. We present methods for summarizing and mining inverse distributions of data streams: they rely on a novel technique to maintain a dynamic sample over the stream with provable guarantees which can be used for variety of summarization tasks (building quantiles or equidepth histograms) and mining (anomaly detection: finding heavy hitters, and measuring the number of rare items), all with provable guarantees on quality of approximations and time/space used by our streaming methods. We also complement our analytical and algorithmic results by presenting an experimental study of the methods over network data streams.

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

Title of host publication | VLDB 2005 - Proceedings of 31st International Conference on Very Large Data Bases |

Pages | 25-36 |

Number of pages | 12 |

State | Published - Dec 1 2005 |

Event | VLDB 2005 - 31st International Conference on Very Large Data Bases - Trondheim, Norway Duration: Aug 30 2005 → Sep 2 2005 |

### Publication series

Name | VLDB 2005 - Proceedings of 31st International Conference on Very Large Data Bases |
---|---|

Volume | 1 |

### Other

Other | VLDB 2005 - 31st International Conference on Very Large Data Bases |
---|---|

Country | Norway |

City | Trondheim |

Period | 8/30/05 → 9/2/05 |

### Fingerprint

### ASJC Scopus subject areas

- Engineering(all)

### Cite this

*VLDB 2005 - Proceedings of 31st International Conference on Very Large Data Bases*(pp. 25-36). (VLDB 2005 - Proceedings of 31st International Conference on Very Large Data Bases; Vol. 1).

**Summarizing and mining inverse distributions on data streams via dynamic inverse sampling.** / Cormode, Graham; Muthukrishnan, Shanmugavelayutham; Rozenbaum, Irina.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*VLDB 2005 - Proceedings of 31st International Conference on Very Large Data Bases.*VLDB 2005 - Proceedings of 31st International Conference on Very Large Data Bases, vol. 1, pp. 25-36, VLDB 2005 - 31st International Conference on Very Large Data Bases, Trondheim, Norway, 8/30/05.

}

TY - GEN

T1 - Summarizing and mining inverse distributions on data streams via dynamic inverse sampling

AU - Cormode, Graham

AU - Muthukrishnan, Shanmugavelayutham

AU - Rozenbaum, Irina

PY - 2005/12/1

Y1 - 2005/12/1

N2 - Emerging data stream management systems approach the challenge of massive data distributions which arrive at high speeds while there is only small storage by summarizing and mining the distributions using samples or sketches. However, data distributions can be "viewed" in different ways. A data stream of integer values can be viewed either as the forward distribution f(x), ie., the number of occurrences of x in the stream, or as its inverse, f -1(i), which is the number of items that appear i times. While both such "views" are equivalent in stored data systems, over data streams that entail approximations, they may be significantly different. In other words, samples and sketches developed for the forward distribution may be ineffective for summarizing or mining the inverse distribution. Yet, many applications such as IP traffic monitoring naturally rely on mining inverse distributions. We formalize the problems of managing and mining inverse distributions and show provable differences between summarizing the forward distribution vs the inverse distribution. We present methods for summarizing and mining inverse distributions of data streams: they rely on a novel technique to maintain a dynamic sample over the stream with provable guarantees which can be used for variety of summarization tasks (building quantiles or equidepth histograms) and mining (anomaly detection: finding heavy hitters, and measuring the number of rare items), all with provable guarantees on quality of approximations and time/space used by our streaming methods. We also complement our analytical and algorithmic results by presenting an experimental study of the methods over network data streams.

AB - Emerging data stream management systems approach the challenge of massive data distributions which arrive at high speeds while there is only small storage by summarizing and mining the distributions using samples or sketches. However, data distributions can be "viewed" in different ways. A data stream of integer values can be viewed either as the forward distribution f(x), ie., the number of occurrences of x in the stream, or as its inverse, f -1(i), which is the number of items that appear i times. While both such "views" are equivalent in stored data systems, over data streams that entail approximations, they may be significantly different. In other words, samples and sketches developed for the forward distribution may be ineffective for summarizing or mining the inverse distribution. Yet, many applications such as IP traffic monitoring naturally rely on mining inverse distributions. We formalize the problems of managing and mining inverse distributions and show provable differences between summarizing the forward distribution vs the inverse distribution. We present methods for summarizing and mining inverse distributions of data streams: they rely on a novel technique to maintain a dynamic sample over the stream with provable guarantees which can be used for variety of summarization tasks (building quantiles or equidepth histograms) and mining (anomaly detection: finding heavy hitters, and measuring the number of rare items), all with provable guarantees on quality of approximations and time/space used by our streaming methods. We also complement our analytical and algorithmic results by presenting an experimental study of the methods over network data streams.

UR - http://www.scopus.com/inward/record.url?scp=33745615174&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33745615174&partnerID=8YFLogxK

M3 - Conference contribution

SN - 1595931546

SN - 9781595931542

T3 - VLDB 2005 - Proceedings of 31st International Conference on Very Large Data Bases

SP - 25

EP - 36

BT - VLDB 2005 - Proceedings of 31st International Conference on Very Large Data Bases

ER -