Relation Extraction is the process of identifying associations between elements in unstructured data. For example, recognizing that Diffbot.com is the URL for the company Diffbot Technologies Corp. Once the relationship is established, data is extracted and structured for consumption in a knowledge graph.
In short, relationship extraction is the process of recognizing the potential for knowledge graph entities and properties in unstructured data. Then pulling out these relationships and structuring them.
Unlike many knowledge bases, knowledge graphs are distinct in that they are often organized to highlight and preserve the relationships between entities. While entity data can be valuable on its own, data on how entities relate to each other can be even more valuable.