search

johnwdubois / rezonator

Rezonator: Dynamics of human engagement
35 stars 2 forks source link

Import GPT #1367

Open johnwdubois opened 1 year ago

johnwdubois commented 1 year ago
  1. Import text from ChatGPT.
  2. Parse the text to identify Participant labels, etc., similar to importing a Transcript #1370
  3. In the Import screen, show an option to import "ChatGPT etc."
  4. Place the new ChatGPT option near the top of the Import screen, reordering the tabs in this order:
    • Transcription
    • AI Chat (e.g. ChatGPT)
    • Song & Verse
    • Interlinear Glossed Text
    • CoNLL-U
    • One Word Per Line (OWPL)
    • Elan (tab-delimited)
    • Prose
  5. In the "Description" field for "ChatGPT etc.", show the following bullet points:
    • "Data from live AI chat session (with ChatGPT)"
    • "Secret API key supplied by user"
    • "Participant labels from “:” (colon)"
    • "Split words on whitespace"
    • "One punctuation unit per line (default)"
    • "One sentence per line (option)"
    • "One turn per line (option)"
  6. Retain the structure of the ChatGPT response ("message") in a node of the .rez file, to allow analysis by rezonateR etc.
  7. Translate key aspects of the ChatGPT response ("message") into varaiables that Rezonator can use.
    • "conversation_id" = DocId
    • "message id" = MessageId
    • "role" = RoleName = ? ParticipantName
    • "parts" = MessageText
    • etc.
  8. Example ChatGPT response: { "message": { "id": "5a5e0a89-99d0-4d9c-a8ec-765f5ea26b83", "role": "assistant", "user": null, "create_time": null, "update_time": null, "content": { "content_type": "text", "parts": [ "I am an AI language model and do not have the ability to experience emotions or feelings. However, I am here to assist you with any questions or information you may need. How can I help you today?" ] }, "end_turn": null, "weight": "1.0", "metadata": {}, "recipient": "all" }, "conversation_id": "ab21dc8c-39d4-4589-90b6-ff5c5af364e3", "error": null }

Resources

See also

1370

700

1369

1368

johnwdubois commented 1 year ago
  1. Retain the structure of the ChatGPT response ("message") in a node of the .rez file, to allow analysis by rezonateR etc.
  2. Translate key aspects of the ChatGPT response ("message") into varaiables that Rezonator can use.
    • "conversation_id" = DocId
    • "message id" = MessageId
    • "role" = RoleName = ? ParticipantName
    • "parts" = MessageText
    • etc.