Skip to main content
. Author manuscript; available in PMC: 2019 Oct 1.
Published in final edited form as: IEEE Trans Knowl Data Eng. 2018 Mar 5;30(10):1825–1837. doi: 10.1109/TKDE.2018.2812203

Fig. 5.

Fig. 5.

Overall Performance Evaluation in Different Languages: Precision-recall curves of all methods on three Wikipedia article datasets evaluated by human annotation. The advantages of AutoPhrase over SegPhrase are more significant in non-English languages, especially on the Chinese dataset. It is worth noting that on the Chinese dataset, AutoPhrase outperforms than two popular, pre-trained Chinese phrase extraction models. This firmly demonstrates the ability of AutoPhrase to cross the language barrier.