Today, connecting two entities with a labelled, meaningful relation is a task that is required in such a large variety of situations that there are thousands of papers, research works and approaches about relation identification. Depending on the application using or producing "relations" or "links", this notion covers actually different realities: from a natural language phrase to a formal formula, from a relationship between 2 entities to one between classes, from domain specific to universal relations, from contextual and possible relations to validated and proved links ... Terminology is one of these research and application domains where relations play a major role to structure and organise language. They contribute to account for meaning and illustrate the difficulty to fix a dynamic and moving reality in a static model. In my talk, I shall first draw an overview of these diverse realities behind the notion of relation in research areas close to linguistics, terminology, knowledge modelling, natural language processing and semantic web. Then I shall connect these definitions with tools, methods and approaches to identify these relations. I shall finally comment how terminology can play a key role in this context, either to change its practices by integrating results and tools from other research fields, or by contributing to improve these approaches with the studies carried out during the last 20 years to build terminologies and terminological knowledge bases.