Simple Morpheme Labelling in Unsupervised Morpheme Analysis

This paper presents my participation to the second Morpho Challenge. Results have been obtained with the algorithm already presented at Morpho Challenge 2005. The system takes a plain list of words as input and returns a list of labelled morphemic segments for each word. Morphemic segments are obtained by an unsupervised learning process which can directly be applied to different natural languages. The system first relies on segment predictability within the longest words in the input word list to identify a set of prefixes and suffixes. Stems are then acquired by stripping affixes from the words. In a third step, words sharing a common stem are compared and split in similar and dissimilar parts corresponding to morphemic segments. Finally, the best segmentation is chosen for each word among all possible segmentations. Results obtained at competition 1 (evaluation of the morpheme analyses) are better in English, Finnish and German than in Turkish. For information retrieval (competi...

• Type: Conference
• Year: 2007
• Where: CLEF
• Authors: Delphine Bernhard
• Keywords: CLEF 2007, Information Retrieval, Morphemic Segments, Morpho Challenge, Unsupervised Learning