Data (or knowledge) integration means the combination of data sources so that they may be used together - in particular, so that queries can be written and executed that span multiple sources.

To permit integration, data must be represented in a way that permits links to be forged between them. The best way to forge links is symbolically by identifying that a symbol in one source means what the same symbol means in another source. Then anything said using the symbol in one source can be matched with anything said using the symbol in the other source, and this kind of linking can enable question answering via search and inference.

In the Semantic Web approach, the symbols are called URIs, and the data is represented in RDF. When RDF combines with RDF, the URIs match up (ideally) and the result has the combined meaning of all of the parts.