Health oriented API design for the public conversation
summary
Twitter embeds contextual annotations embedded in v2 payload so NLP named‑entity recognition tags tweets (teams, holidays, topics) rather than relying on huge keyword lists. A COVID-19 filtered stream endpoint returns only tweets the model marks as COVID‑19 across languages, enabling real‑time research without bespoke training. A hide replies API plus the accounts activity webhook and Perspective API scoring lets apps auto‑hide likely‑toxic replies in real time (analysis stays local to the app flow), demonstrably reducing visible abuse.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:35Health oriented API design
- 7:23Context awareness
- 8:57Improving the public conversation
- 12:57Hiding replies
- 14:11Hide replies API
- 15:52Perspective API
- 16:53Integration with Twitter
- 18:14Benefits for internal processes
- 19:39Conclusion








