This paper presents the collaborative participation between the IRIS team and the LIMSI laboratory to the Trilingual Entity Discovery and Linking of TAC KBP 2017. The aim of the EDL track is to evaluate systems that automatically detect entities in raw text and manage to linking them. In our first joint participation, we focused on the entity linking task and use a public available software for the entity recognition task. Results show that an extra effort must be performed to improve the

more »

... recognition phase in order to improve the later entity linking step.