Key summary
Based on publicly usable agreements and the approved Sejong, Buan, and Paju scope, this article distinguishes each participant’s role in research design, device operations, field records, and commercialization decisions. A single concentration value cannot prove reduction. Measurement location, device health, calibration, missing intervals, temperature, humidity, ventilation, and operational events must be reviewed on one timeline with traceable records.
Measurement design and data flow
Define the measurement boundary, sampling frequency, observation period, and device locations first. Record clock alignment, maintenance, configuration changes, and missing intervals; do not present interpolated values as observations. Before-and-after comparisons need predefined windows and exclusion rules, with season, time of day, ventilation, and feeding conditions reviewed.
Interpretation and operator review
AI prioritizes patterns and anomaly candidates; it does not make the final decision. Operators review field logs and device health before acting. The dashboard must update every view when farm or period changes, while reports preserve data quality, method, results, uncertainty, and approval history.
FAQ
Can continuous data alone prove reduction?
No. Baseline and post-application observations must be compared under equivalent conditions, with device health, environment, missing data, and operational events reviewed.
Does AI replace operator judgment?
No. AI helps find patterns and anomaly candidates; actions and report approval still require human review.
Evidence and limitations
Publicly usable agreements; approved Sejong, Buan, and Paju field scope. Patents and agreements describe technical direction and collaboration scope; they do not guarantee a reduction rate, certification, investment, or commercial outcome. Screens and values are demonstrative.