Abstract

Automatic template extraction has been studied intensively in order to perform information extraction without predefined template. Several existing studies utilized the similar preprocessing techniques which are applied in Open Information Extraction (Open IE) paradigm system. We investigate the use of Open IE results to build the automatic event template extraction. In this study, we adapt the clustering based approach for template extraction, and propose to add the relation similarity information in the clustering function. We compare the clusters quality of the Open IE based system and non-Open IE based system and also with the use of relation similarity function using document classification metric. The experimental result shows that the performance of Open IE based system is comparable with the non-Open IE based system and the relation similarity information is able to improve the clusters quality.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call