Stage 1 — Data foundation
Australian Oncology X Social Listening
The underlying data structure is in place, with the three seed directories loaded. Discovery, analysis and weekly reporting are deliberately not built yet.
Directory
- Entities — people and organisations, held independently of any X account
- Classifications — one entity can hold several categories at once
- Tumour and cancer areas per entity
- X accounts — one canonical record per account, no duplicates
- Verification history — a bounded record of each attempt and its outcome
Capture
- Posts — original, repost, quote and reply, with publication date kept separate from capture date
- Conversations — grouping so activity is judged per discussion, not per post
- Activity per conversation per week, separating where a discussion began from where it continued
- Discovery records — where each candidate came from
- Topic terms, including terms that emerge later
Reporting and retention
- Reporting periods — Friday to Thursday in Australian Eastern local time, daylight saving included
- Themes and distinct viewpoints, always tied to a period
- Weekly reports — snapshots that cannot change once finalised
- Report items — the ranked contents, each with its detail frozen at generation time
- Retention tiers and lightweight historical conversation records
Open the weekly discovery run and its checks
Open the automated analysis of captured posts
Paste a real X sample into an isolated test week
Account rule held in the structure
An entity can exist with no X account, an unconfirmed account, or a confirmed one. Only a confirmed, active, deliberately seeded account is eligible for monitoring — everything else stays in the directory for recognition and later verification without being searched for again on every cycle.