Creating the First Dataset

[akraradets]

Requirement

1. 500 - 600 words
2. Topic: Telehealth

• Doctor–patient consultation (symptoms, diagnosis, treatment). This can be too broad for us.
• Appointment booking and Medication explanation and side effects.
• We can collect common phrases/words from like; Hospital forms (registration, consent, medical history), Health information pages (symptoms, diseases, treatments).
• Extract all words and keep only meaningful content words (nouns, verbs, adjectives, important phrases).
• We need doctors, nurses, and signers for validation.

Resources

SignBank (SignPuddle)

[akraradets]

@salmiakki0512 @phuepwint-thwe

• Generate 4 - 5 conversations doctor-patient
• Count unique words
• Lookup the word against the SignBank.
• Arrange a meeting with the Signer on how to organize the dataset for recording
• Have a conversation with Dr. Chantri on using the studio

[phuepwint-thwe]

To do list for the signer meeting

• check the unique words on previous work(ASL and German Dataset)
• Vector based glossing (Gloss2Vec)
• Can we map the moving sequences into a vector?
• List down the questions and concerns of all of us before the meeting with Signer
• Document on the generation of conversation. ( like context metter )
• Define the context.
• How many context we need to consider? Can we be replace by only text? Can we only focus on the general conversation?