Automatic Extraction of Arabic Multiword Expressions

摘要

In this paper we investigate the automatic acquisition of Arabic Multiword Expressions (MWE). We propose three complementary approaches to extract MWEs from available data resources. The onrst approach relies on the correspondence asymmetries between Arabic Wikipedia titles and titles in 21 different languages. The second approach collects English MWEs from Princeton WordNet 3.0, translates the collection into Arabic using Google Translate, and utilizes different search engines to validate the output. The third uses lexical association measures to extract MWEs from a large unannotated corpus. We experimentally explore the feasibility of each approach and measure the quality and coverage of the output against gold standards.