Sequence Segmentation by Enumeration: An Exploration
Steffen Eger · The Prague Bulletin of Mathematical Linguistics · 2013
Abstract We investigate exhaustive enumeration and subsequent language model evaluation (E&E approach) as an alternative to solving the sequence segmentation problem. We show that, under certain conditions (on string lengths and regarding a possibility to accurately estimate the number of segments), which are satisfied for important NLP applications, such as phonological segmentation, syllabification, and morphological segmentation, the E&E approach is feasible and promises superior results than the standard sequence labeling approach to sequence segmentation.