Table 21. Web-KB – significance test of BLAB-SM against all measures.

Measure	MAE	MSE	Std error	Confidence interval
Manhattan	0.602047015	0.362962271	0.002891546	[0.39216701–0.40373895]
Euclidean	0.239183835	0.058129364	0.003916753	[0.75297875–0.76865356]
Cosine	0.238087691	0.057424802	0.003509637	[0.75488953–0.76893507]
Jaccard	0.776096143	0.602833150	0.002909542	[0.21808187–0.22972583]
KLD	0.777548864	0.605136971	0.003040655	[0.21636679–0.22853547]
SMTP	0.195668251	0.038413104	0.001455103	[0.80142009–0.80724340]
Ex-Jaccard	0.240649762	0.058737734	0.003709056	[0.75192843–0.76677204]
BLAB-SM	0.186283068	0.034801849	0.001294012	[0.81112761–0.81630624]

Note:

The smaller the values of MAE, MSE and Std error, the better the similarity measure is. Bolded numerical values in the BLAB-SM, SMTP and Ex-Jaccard rows are the best measure whose values of MAE, MSE and Std error are the smallest.