Mining Class-Correlated Patterns for Sequence Labeling

13 years 3 months ago

Download wwwkramer.in.tum.de

Sequence labeling is the task of assigning a label sequence to an observation sequence. Since many methods to solve this problem depend on the speciﬁcation of predictive features, automated methods for their derivation are desirable. Unlike in other areas of pattern-based classiﬁcation, however, no algorithm to directly mine class-correlated patterns for sequence labeling has been proposed so far. We introduce the novel task of mining class-correlated sequence patterns for sequence labeling and present a supervised pattern growth algorithm to ﬁnd all patterns in a set of observation sequences, which correlate with the assignment of a ﬁxed sequence label no less than a user-speciﬁed minimum correlation constraint. From the resulting set of patterns, features for a variety of classiﬁers can be obtained in a straightforward manner. The eﬃciency of the approach and the inﬂuence of important parameters are shown in experiments on several biological datasets. Key words: seque...

Thomas Hopf, Stefan Kramer

Real-time Traffic