Relation extraction has been considered as one of the most popular topics nowadays, thanks for its common application in knowledge graph, machine reading and other artificial intelligence sub-field. However, this field has long been suffered from data hunger. Annotating large high-quality datasets for relation extraction is troublesome and time-consuming. This thesis project will main focus on efficient way of annotating text datasets for extracting complex relations between entities. Moreover, we put some efforts on compare the influence of different components in the pipeline. The main contributions of this project are the comparisons and analysis regarding the influences of components, which are in place for the majority of relation extraction models, and the clear literature review together with the summary of available datasets in the relation extraction flow.