Transfer Building of Multiword Expression Resource from Indonesian to Malay
Wuying Liu, Lin Wang · 2020
The scarcity of corpus makes multiword expression (MWE) extraction very difficult for some low-resource languages. The homology and similarity between languages bring a new opportunity about morphological transfer learning from rich-resource languages to low-resource languages. This paper addresses the issue of MWE extraction for relatively low-resource Malay, and proposes a novel transfer building algorithm from rich-resource Indonesian. The experimental results verify the effectiveness of our algorithm to extract Malay MWEs from Indonesian corpus, and also prove a strong computability of morphological transfer learning from Indonesian to Malay owing to the homology and similarity of the same Malay-Polynesian language branch of the Austronesian language family.