In a knowledge base, entities are considered as stable but some events can break relations. This is the case with events involving organizations, products, or brands, which can be bought. In this article, our approach is aimed at extracting relations of type "acquisition" between two entities. Relation extraction from a dynamic source of information such as Twitter enables the detection of entities in real time. This, based on events detection, allows updating a database accordingly. The approach consists in modeling events based on a lexico-semantic resource. Then, once the entities are linked to the Linked Open Data, linguistic rules are applied, finally, to generate RDF triples.