A Topic Recognition Method of News Text Based on Word Embedding Enhancement

Table 2

The information of the 20NewsGroup dataset.

Topic category	Len	Size	Topic category	Len	Size
comp.graphics	163.0	973	rec.autos	126.3	990
comp.os.ms.windows.misc	160.6	985	rec.motorcycles	118.4	996
comp.sys.ibm.pc.hardware	116.7	982	rec.sport.baseball	131.2	994
comp.sys.mac.hardware	109.0	963	rec.sport.hockey	155.3	999
comp.windows.x	174.3	988	misc.forsale	95.1	975
talk.politics.misc	232.7	775	sci.crypt	189.6	991
talk.politics.guns	189.9	910	sci.electronics	121.8	984
talk.politics.mideast	269.4	940	sci.med	173.8	990
alt.atheism	183.0	799	sci.space	174.9	987
soc.religion.Christian	194.2	997	talk.religion.misc	192.1	628