Back to Projects
Pipelinev0.8

CNT X-ray Market Newsletter

A daily newsletter that gathers and filters X-ray news, papers and patents

9
Sources
2-stage
Filtering
0
Human steps before sending

Highlights

Two-stage filtering

Keywords make the first cut and judgment rules the second; the rules are tuned by backtesting against a labeled set

Cut-log feedback

Dropped items are logged, and missed keywords found there go back into the rules

Unattended

Every day at a set time it runs from collection to sending with no one touching it

Pipeline

  1. 1
    Collect
    News · papers · patents · filings
  2. 2
    First cut
    Keywords
  3. 3
    Second cut
    Judgment rules
  4. 4
    Dedupe
    SQLite
  5. 5
    Send
    HTML email

Stack

PythonSQLiteGitHub Actions