Masayu Leylia Khodra
STE-ITB
M. Abdi Haryadi. H
abdiharyadi.ah@gmail.com
Text style transfer (TST) is an NLP task that transforms a specific style while preserving the main content of the input text. AMR-TST (Shi et al., 2023) is a TST method that utilizes an abstract meaning representation (AMR) graph to preserve semantics. This method can be applied to Indonesian by utilizing sentiment triplet extraction (William and Khodra, 2022) to detect stylistic words.
Referensi
Alzamzami, MN (2023). Development of a Cross-Language Abstract Meaning Representation Parser between Indonesian and English with BART, Input Concatenation, and Data Augmentation (Master's Thesis). Bandung: Bandung Institute of Technology. Shi, K. et al. (2023). AMR-TST: Abstract Meaning Representation-Based Text Style Transfer. ACL 2023.
William & Khodra, M.L. (2022). Generative Opinion Triplet Extraction Using Pretrained Language Model. IEEE.
Daryanto, T.H. & Khodra, M.L. (2022). Indonesian AMR-to-Text Generation by Language Model Fine-tuning. IEEE.