The Project — Research & NLP

The research

Avian Flu Watch is the public face of ongoing research into how H5N1 avian influenza spreads through Australian wildlife, and how early signals of that spread can be detected and mapped faster than traditional reporting alone allows.

What we are trying to achieve with NLP

Official surveillance data is authoritative but slow: a detection can circulate in local news, community groups, and social media days before it appears in a government dataset. We use natural language processing to close that gap.

• News articles, bulletins, and community-submitted links are read by an AI extraction pipeline that pulls out structured detection candidates — species, location, date, count, and whether a mass-mortality event is described.
• Every extracted candidate is reviewed by a human moderator before it appears on the map, and is always labelled as unconfirmed until officially verified.
• The goal: a live national picture that leads official reporting by days, without ever presenting unverified information as confirmed.