The Markov Expert for Finding Episodes in Time Series

We describe a domain-independent, unsupervised algorithm for refined segmentation of time series data into meaningful episodes, focusing on the problem of text segmentation. The VOTING EXPERTS algorithm of Cohen et al. [1] achieves results with fairly low rates of error. The MARKOV EXPERT is a new approach that improves the performance of VOTING EXPERTS by further refining those results with votes from an additional expert. The new expert applies a Markov-based segmentation method inspired by the approach of Teahan et al. [2], using the results of VOTING EXPERTS' frequency and entropy experts as a sample corpus from which to draw prefix/suffix frequency data. In contrast with the setting of Teahan et al., in this setting the sample corpus is small and somewhat inaccurate, but despite its errors, it is directly similar to the intended output in terms of non-space characters. The result is a high quality domainindependent segmentation algorithm that performs substantially better th...

• Type: Conference
• Year: 2005
• Where: DCC
• Authors: Jimming Cheng, Michael Mitzenmacher
• Keywords: Computer Graphics, DCC 2005, Domainindependent Segmentation Algorithm, Markov-based Segmentation Method, VOTING EXPERTS Algorithm