Table 5 Descriptive measures: total...

Table 5

Descriptive measures: total number of matches, average similarity, standard deviation, and number of matches by high, medium and low confidence similarity of the comparison at unit level of SOC and ISCO are detailed through the methods BoW, TF-IDF, SBERT and Ensemble

	BoW	TF-IDF	SBERT	Ensemble
Matches found	421	264	1,308	1,096
Average similarity	50.22%	52.36%	68.39%	47.75%
Standard deviation	19.27%	22.45%	13.25%	14.78%
High confidence	38	39	238	52
Medium confidence	65	31	713	108
Low confidence	318	194	357	936

	BoW	TF-IDF	SBERT	Ensemble
Matches found	421	264	1,308	1,096
Average similarity	50.22%	52.36%	68.39%	47.75%
Standard deviation	19.27%	22.45%	13.25%	14.78%
High confidence	38	39	238	52
Medium confidence	65	31	713	108
Low confidence	318	194	357	936

[ViewLarge]

Sharing Unavailable