Data joining combines records or collections from multiple sources into a connected dataset. It can enrich a raw or proprietary dataset with labels, public information, behavioral profiles, or another modality. section 6.1
Choose the joining pattern according to the data and the business goal:
- Use a straightforward connection when datasets share comparable records, such as a SQL join between tables. section 6.1
- Use multimodal joining when different kinds of information describe the same problem, such as product images combined with text descriptions. section 6.1
- Check access and legality before combining data, especially when exchanges cross organizations or regions. section 6.1
Game publisher and ad network
A game publisher keeps its paying-customer list private, while a partner ad network holds detailed behavioral profiles. Joining the datasets can help target advertising toward users resembling paying players, creating value for both businesses. section 6.1
Figure 6.1 Combining data sources of two completely different businesses can eventually benefit both.
Trace how two businesses combine distinct data sources to create a shared marketing opportunity. section 6.1
