Amazon Translate

Amazon Translate 101

🔧 Text translation service
- Powered by neural networks
- Translates text to other languages, one word at a time
Translation happens in two parts:
1. Encoder reads source → outputs semantic representation (i.e. meaning of source text)
  - 💡 A good translation doesn't directly translate word for word, the whole meaning and context needs to be conveyed
2. Decoder reads meaning → writes target language
Features
- Attention mechanisms help understand context, ensure “meaning” is translated
  - Identify which words in source text are most relevant for context and generating target output
- Can auto detect source text language
- Can use terminologies, i.e. custom dictionaries with your own custom translations for specific terms
- Parallel data → several possible translations depending on context
  - e.g. ENG: “how are you?” → FR: “Comment ça va?” or “Comment allez-vous?”, depending on situation
Use cases
- Offer multilingual UX
  - e.g. meeting notes, posts, communications, articles…
  - e.g. emails, in-game chat, customer live chat
- Translate incoming data (social media, news, communications…)
- Common use case: integrate with other services (AWS or external), apps and platforms to offer language-independence
  - e.g. Comprehend, Transcribe, Polly…
  - Can translate data stored in S3, RDS, DDB…
Screenshot