Name
Abstract | ||
|---|---|---|
Sequential patterns mining is an important data mining approach with broad applications. Traditional mining algorithms on database were not adapted to data stream. Recently, some approximate sequential pattern mining algorithms over data stream were presented which solved some problems except the one of wasting too many system resources in processing long sequences. According to observation and proof, a novel approximate sequential pattern mining algorithm is proposed named GraSeq. GraSequses directed weighted graph structure and stores the synopsis of sequences with only one scan of data stream; furthermore, a subsequences matching method is mentioned to reduce the cost of long sequences' processing and a conception validnodeis introduced to improve the accuracy of mining results. Our experimental results demonstrate that this algorithm is effective and efficient. |
Year | DOI | Venue |
|---|---|---|
2007 | 10.1007/978-3-540-73871-8_37 | ADMA |
Keywords | Field | DocType |
data stream,novel approximate sequential pattern,approximate sequential pattern mining,important data mining approach,traditional mining algorithm,sequential patterns,long sequence,sequential patterns mining,mining result,novel approximate mining approach,mining algorithm,broad application,sequential pattern mining,data mining | Data mining,Graph,Data stream mining,Pattern recognition,Data stream,Computer science,Artificial intelligence,Sequential Pattern Mining,Machine learning | Conference |
Volume | ISSN | Citations |
4632 | 0302-9743 | 6 |
PageRank | References | Authors |
0.56 | 9 | 2 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Haifeng Li | 1 | 46 | 6.19 |
| Hong Chen | 2 | 99 | 23.20 |