Research
A real system for a question that cannot be studied in isolation.
AdPrecision is a personal, non-commercial project, developed in the course of research on recommender systems for public and digital-out-of-home displays. It is not sold, licensed or monetised.
The research question
Recommender systems for public and DOOH displays: for groups of anonymous, momentary viewers rather than known individuals, and for requests that must satisfy hard commercial constraints rather than being ranked by preference alone.
The model
Why the commercial side is so detailed.
Bookings and guarantees define promises. Capacity and delivery gaps define constraints. Proof of play and yield provide measurable outcomes.
Together they turn “pick a good ad” into a problem with an objective and hard limits, one that can be explained, evaluated and repaired.
ObjectiveRelevant delivery
PromisesBookings & guarantees
ConstraintsCapacity & availability
EvidenceProof of play & yield
Status
What ships, and what is direction.
Ships today
- The full path from a file to a measured second of airtime: authoring, scheduling, selling, delivery, measurement.
- Bookings with tiers and guarantees, and a fit and repair check against airtime capacity.
- Proof of play per spot, sealed hourly into a verifiable chain.
Research direction
- A knowledge-based recommender: multi-attribute utility scoring for audience-level relevance.
- Constraint diagnosis in the tradition of QuickXplain and FastDiag, to explain and repair requests that do not fit.
- Measured delivery as the reward signal, not an assumed one.
Who builds it: the responsible person is named in the imprint.
Read the research context.
The documentation carries the detail, from the fit and repair engine to the proof chain.