A clustering-based topic model using word networks and word embeddings

Mu, Wenchuan; Lim, Kwan Hui; Liu, Junhua; Karunasekera, Shanika; Falzon, Lucia; Harwood, Aaron

doi:10.1186/s40537-022-00585-4

Journal of Big Data

Table 8 Comparison of ClusTop algorithm against various baselines, in terms of Precision (Pre), Recall (Rec) and F-score (FS) for the top 15 keywords/unigrams of each topic

From: A clustering-based topic model using word networks and word embeddings

Algorithm	Top 15 Keywords/Unigrams
	Dataset A			Dataset B			Dataset C			Average
	Precision	Recall	F-score	Precision	Recall	F-score	Precision	Recall	F-score	Rank@15
ClusTop-Word-NA	.648 ± .002 (24)	.034 ± .000 (5)	.063 ± .000 (7)	.743 ± .004 (23)	.053 ± .000 (3)	.098 ± .001 (3)	.717 ± .010 (22)	.038 ± .001 (4)	.070 ± .002 (4)	(10.6)
ClusTop-BiG-NA	.671 ± .002 (21)	.035 ± .000 (2)	.065 ± .000 (2)	.747 ± .004 (21)	.055 ± .001 (1)	.100 ± .001 (1)	.719 ± .010 (21)	.040 ± .001 (2)	.073 ± .002 (2)	(8.1)
ClusTop-TriG-NA	.667 ± .002 (22)	.035 ± .000 (2)	.065 ± .000 (2)	.757 ± .004 (20)	.055 ± .000 (1)	.100 ± .001 (1)	.677 ± .010 (24)	.041 ± .001 (1)	.075 ± .002 (1)	(8.2)
ClusTop-BiHa-NA	.667 ± .002 (22)	.034 ± .000 (5)	.064 ± .000 (5)	.729 ± .004 (24)	.052 ± .000 (4)	.096 ± .001 (4)	.722 ± .009 (20)	.040 ± .001 (2)	.073 ± .002 (2)	(9.8)
ClusTop-Hash-NA	.848 ± .004 (9)	.023 ± .000 (24)	.045 ± .000 (18)	.880 ± .008 (9)	.031 ± .001 (17)	.060 ± .001 (17)	.914 ± .012 (8)	.023 ± .001 (17)	.044 ± .001 (17)	(15.1)
ClusTop-Noun-NA	.698 ± .002 (18)	.033 ± .000 (8)	.061 ± .000 (9)	.773 ± .003 (19)	.051 ± .000 (5)	.093 ± .001 (5)	.753 ± .010 (15)	.036 ± .001 (6)	.067 ± .001 (6)	(10.1)
ClusTop-H2VG-NA	.895 ± .004 (2)	.024 ± .000 (23)	.046 ± .000 (17)	.961 ± .005 (1)	.031 ± .001 (17)	.060 ± .001 (17)	.962 ± .008 (3)	.023 ± .001 (17)	.044 ± .001 (17)	(12.7)
ClusTop-H2VW-NA	.889 ± .001 (5)	.026 ± .001 (21)	.026 ± .001 (24)	.935 ± .001 (3)	.030 ± .001 (20)	.030 ± .001 (20)	.983 ± .001 (1)	.020 ± .001 (23)	.020 ± .001 (23)	(15.6)
ClusTop-H2VF-NA	.862 ± .001 (7)	.028 ± .001 (14)	.028 ± .001 (19)	.903 ± .001 (8)	.030 ± .001 (20)	.030 ± .001 (20)	.942 ± .001 (6)	.021 ± .001 (20)	.021 ± .001 (20)	(14.9)
ClusTop-Word-AH	.677 ± .002 (20)	.025 ± .000 (22)	.048 ± .000 (16)	.744 ± .004 (22)	.043 ± .000 (12)	.079 ± .001 (14)	.731 ± .010 (19)	.033 ± .001 (7)	.061 ± .001 (8)	(15.6)
ClusTop-Hash-AH	.802 ± .002 (10)	.030 ± .000 (12)	.058 ± .000 (12)	.831 ± .004 (13)	.044 ± .000 (10)	.083 ± .001 (10)	.847 ± .009 (12)	.026 ± .001 (15)	.050 ± .001 (15)	(12.1)
ClusTop-Noun-AH	.699 ± .002 (17)	.027 ± .000 (16)	.051 ± .000 (14)	.791 ± .004 (16)	.042 ± .000 (14)	.078 ± .001 (15)	.781 ± .010 (13)	.033 ± .001 (7)	.062 ± .001 (7)	(13.2)
ClusTop-H2VG-AH	.762 ± .002 (12)	.029 ± .000 (13)	.055 ± .000 (13)	.832 ± .003 (12)	.043 ± .000 (12)	.080 ± .001 (12)	.859 ± .009 (11)	.028 ± .001 (10)	.053 ± .001 (11)	(11.8)
ClusTop-H2VW-AH	.902 ± .001 (1)	.027 ± .001 (18)	.027 ± .001 (22)	.913 ± .001 (5)	.029 ± .001 (22)	.029 ± .001 (22)	.964 ± .001 (2)	.020 ± .001 (23)	.020 ± .001 (23)	(15.3)
ClusTop-H2VF-AH	.893 ± .001 (3)	.028 ± .001 (14)	.028 ± .001 (19)	.913 ± .001 (5)	.030 ± .001 (20)	.030 ± .001 (20)	.945 ± .001 (5)	.021 ± .001 (20)	.021 ± .001 (20)	(14.0)
ClusTop-Word-AM	.687 ± .001 (19)	.036 ± .000 (1)	.067 ± .000 (1)	.790 ± .004 (17)	.046 ± .000 (7)	.085 ± .001 (7)	.689 ± .012 (23)	.027 ± .001 (13)	.052 ± .001 (13)	(11.2)
ClusTop-Hash-AM	.713 ± .002 (15)	.033 ± .000 (8)	.063 ± .000 (7)	.816 ± .004 (14)	.045 ± .000 (9)	.083 ± .001 (10)	.770 ± .011 (14)	.030 ± .001 (9)	.058 ± .001 (9)	(10.6)
ClusTop-Noun-AM	.860 ± .001 (8)	.026 ± .000 (20)	.050 ± .000 (15)	.905 ± .003 (7)	.041 ± .000 (16)	.077 ± .001 (16)	.892 ± .008 (9)	.026 ± .000 (15)	.049 ± .001 (16)	(13.6)
ClusTop-H2VG-AM	.758 ± .002 (13)	.031 ± .000 (11)	.060 ± .000 (10)	.863 ± .003 (10)	.044 ± .000 (10)	.083 ± .001 (10)	.876 ± .008 (10)	.027 ± .000 (13)	.052 ± .001 (13)	(11.1)
ClusTop-H2VW-AM	.892 ± .001 (4)	.027 ± .001 (18)	.027 ± .001 (22)	.942 ± .001 (2)	.028 ± .001 (23)	.028 ± .001 (23)	.960 ± .001 (4)	.021 ± .001 (20)	.021 ± .001 (20)	(15.1)
ClusTop-H2VF-AM	.877 ± .001 (6)	.027 ± .001 (18)	.027 ± .001 (22)	.935 ± .001 (3)	.027 ± .001 (24)	.027 ± .001 (24)	.931 ± .001 (7)	.021 ± .001 (20)	.021 ± .001 (20)	(16.0)
LDA-Orig	.799 ± .002 (11)	.032 ± .000 (10)	.060 ± .000 (10)	.840 ± .003 (11)	.042 ± .000 (14)	.080 ± .001 (12)	.753 ± .010 (15)	.028 ± .001 (10)	.054 ± .001 (10)	(11.4)
LDA-Hash	.720 ± .002 (14)	.034 ± .000 (5)	.064 ± .000 (5)	.795 ± .003 (15)	.046 ± .000 (7)	.085 ± .001 (7)	.733 ± .010 (18)	.038 ± .001 (4)	.070 ± .001 (4)	(8.8)
LDA-Ment	.703 ± .002 (16)	.034 ± .000 (5)	.064 ± .000 (5)	.774 ± .004 (18)	.048 ± .000 (6)	.088 ± .001 (6)	.743 ± .012 (17)	.027 ± .001 (13)	.052 ± .001 (13)	(11.0)

The rank of an algorithm’s performance for each metric are provided in brackets

Back to article page