Order from Lab F930E0

Media interactions

Long, real interaction histories from a music, video or podcast service (plays, skips, ratings, and what was shown but not chosen), joined to the content catalogue, with each user a stable pseudonym. We need whole, untruncated per-user histories with consent for model training, not a sampled or re-keyed public set.

How answering works

Any number of people fill one order. The lab receives one dataset, in its own columns.

Your agent reads the order
It looks at what you hold and says which columns you can fill, and what is missing.
Mora checks what you send
Every record: no copies, no personal data, not already public, not on Mora from someone else.
The lab pays per record it accepts
The lab named the price. You fill as much of the order as you can.

The columns the lab wants

You do not need every column. Mora says which ones each dataset fills.

Column Type What it must hold

user_id id Pseudonymous listener or viewer, the same id for the whole history, not reversible by the seller

session_id id One sitting

item_id id Track, episode, film or show, joined to the items table

event_type label play, skip, pause, seek, like, dislike, rating, add, search

event_at timestamp When it happened, to the second, with timezone offset

ms_played number How long the item was actually played

rating number Explicit rating when the user gave one

surface label Where the play came from: home, search, autoplay, playlist, recommended slot

position number Rank of the item in the list it was chosen from

device_type label phone, tv, web, speaker

impressions table Items shown and not chosen: user_id, session_id, shown_at, item_id, position, surface, acted

item metadata table item_id, kind, title, creator_id, duration_ms, genres, language, released_at, added_at

user attributes table Coarse only: signup_month, country, plan, age_band

consent document Policy version, period in force, lawful basis, opt-outs removed

Column	Type	What it must hold
`user_id`	id	Pseudonymous listener or viewer, the same id for the whole history, not reversible by the seller
`session_id`	id	One sitting
`item_id`	id	Track, episode, film or show, joined to the items table
`event_type`	label	play, skip, pause, seek, like, dislike, rating, add, search
`event_at`	timestamp	When it happened, to the second, with timezone offset
`ms_played`	number	How long the item was actually played
`rating`	number	Explicit rating when the user gave one
`surface`	label	Where the play came from: home, search, autoplay, playlist, recommended slot
`position`	number	Rank of the item in the list it was chosen from
`device_type`	label	phone, tv, web, speaker
`impressions`	table	Items shown and not chosen: user_id, session_id, shown_at, item_id, position, surface, acted
`item metadata`	table	item_id, kind, title, creator_id, duration_ms, genres, language, released_at, added_at
`user attributes`	table	Coarse only: signup_month, country, plan, age_band
`consent`	document	Policy version, period in force, lawful basis, opt-outs removed

How much, in the lab's words

At least 1 million users, each with 12 months or more of unbroken history; about 3 billion events; 90% of events joined to an item row. Parquet, split by day.

events
items
impressions
users
consent