Raedt. Constraint Based Mining of First Order Sequences in SeqLog (2004)
| Venue: | Database Support for Data Mining Application |
| Citations: | 11 - 0 self |
BibTeX
@INPROCEEDINGS{Lee04raedt.constraint,
author = {Sau Dan Lee and Luc De Raedt},
title = {Raedt. Constraint Based Mining of First Order Sequences in SeqLog},
booktitle = {Database Support for Data Mining Application},
year = {2004},
pages = {155--176},
publisher = {Springer}
}
Bookmark
 |
 |
 |
 |
 |
 |
OpenURL
Abstract
Abstract. A logical language, SeqLog, for mining and querying sequential data and databases is presented. In SeqLog, data takes the form of a sequence of logical atoms, background knowledge can be specified using Datalog style clauses and sequential queries or patterns correspond to subsequences of logical atoms. SeqLog is then used as the representation language for the inductive database mining system MineSeqLog. Inductive queries in MineSeqLog take the form of a conjunction of a monotonic and an anti-monotonic constraint on sequential patterns. Given such an inductive query, MineSeqLog will compute the borders of the solution space. MineSeqLog uses variants of the famous level-wise algorithm together with ideas from version spaces to realize this. Finally, we report on a number of experiments in the domains of user-modelling that validate the approach. 1
Citations
No citations identified.
