Data that isn't internally managed database exports.

One of the reasons big data systems took off was because enterprises had exports out of third party systems that they didn't want to model since they didn't own it. As well as a bunch of unstructured data e.g. floor plans, images, logs, telemetry etc.

the other comments get it.

It means that data comes in a ton of different shapes with poorly described schemas (technically and semantically).

From the typical CSV export out of an ERP system to a proprietary message format from your own custom embedded device software.

not op, but I think they mean the data is complex, heterogeneous and noisy. You won't be able to extract meaning trivially from it, you need something to find the (hidden) meaning in the data.

So AI currently, probably ;)

it doesn't fit the relational model, e.g. you have some tables, but also tons of different types of images, video, sounds, raw text, etc.