Paper: The UMUS System for Named Entity Generation at GREC 2010

ACL ID W10-4230
Title The UMUS System for Named Entity Generation at GREC 2010
Venue International Conference on Natural Language Generation
Session Main Conference
Year 2010
Authors

We present the UMUS (Universit´e du Maine/Universit¨at Stuttgart) submission for the NEG task at GREC’10. We re- fined and tuned our 2009 system but we still rely on predicting generic labels and then choosing from the list of expressions that match those labels. We handled recur- sive expressions with care by generating specific labels for all the possible embed- dings. The resulting system performs at a type accuracy of 0.84 an a string accuracy of 0.81 on the development set.