Improving Entity Linking Performance Using Frame Semantics

In this paper, we investigate the use of Frame Semantics for improving the performance of linking of non-PLO (Person, Location, Organization) entities to Wikipedia articles. We propose an architecture for a system using Frame Semantics and Selectional Preferences (SP) to improve precision of linking non-PLO entities. We present performance results, which suggest that our system provides a mechanism to improve entity linking for non-PLO entities. We also show that Selectional Preferences can contribute for improved entity linking.