ANALYSIS OF A SERIES OF MACHINE TRANSLATION MODELS WITH TRANSFORMER FOR MINANGKABAU TO GERMANY
Machine translation has evolved from a word-for-word translation to a complex model that can take the context of a sentence and interpret it more accurately. In this final assignment, the author analyzes a series of machine translation models. The translation machine model used is the Transformer...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/77759 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Machine translation has evolved from a word-for-word translation to a complex
model that can take the context of a sentence and interpret it more accurately. In this
final assignment, the author analyzes a series of machine translation models. The
translation machine model used is the Transformer model. In this final assignment
we will see how the performance of a series of Transformer models is to translate
a series of languages, namely Minangkabu-Indonesia-English-German. In this final
project, the BLEU score is used to assess model performance on test data. The
analysis results show a decrease in the BLEU score for longer translation sequences.
The BLEU score for a 4-language set is smaller than for a set with fewer languages.
The analysis results also show that the Transformer model is not good enough to
translate a language with a simple grammar into a language with a more complex
grammar |
|---|