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. 9.

Fig. 9.

Comparison Between Phrase Mining Methods with/without POS tags (AutoPhrase and AutoSegPhrase) as input. Datasets in different languages are used for the comparison. The best baseline in each dataset is provided as a reference. The results show POS-guided phrasal segmentation works more smoothly in different languages. As the original segmentation method without POS information is designed for English, it works well for English and Spanish but relatively poorly on the Chinese data.