Anticipating Corporate Financial Performance from CEO Letters Utilizing Sentiment Analysis

Table 2

Related work of machine-learning approach in sentiment analysis.

Authors	Objectives	Models	Data source	Evaluation method	Data set	Accuracy (percent)	Precision (percent)	Recall	F1
Bruce and Wiebe [33]	Assign sentences subjectivity using 1 to 4 scale	LCA	14 articles in Wall Street Journal Treebank Corpus	10-fold cross validation	486 subjective, 515 objective	72.17	N/A	N/A	N/A
Pang et al. [30]	Assign docs sentiments	SVM, NB, ME	Movie reviews	3-fold cross validation	700 (+)	77–82.9	N/A	N/A	N/A
					700 (−)
Bo and Lee [1]	Assign docs sentiments	SVM, NB	Movie reviews	10-fold cross validation	1000 (+)	86.4–87.2	N/A	N/A	N/A
					1000 (−)
Gamon and Michael [34]	Assign docs sentiments using 4-point scale	SVM	Customer feedback	10-fold cross validation	N/A	77.5	N/A	N/A	N/A
				10-fold cross validation	N/A	69.5	N/A	N/A	N/A
Yang et al. [35]	Assign topic sentiments	Class association rules	Consumer reviews	Macroaveraged	N/A	N/A	67.64	77.5	72.26
				Microaveraged (microaveraging means given a set of confusion tables, a new two-by-two contingency table is generated, each cell in the new table represents the sum of the number of documents from within the set of tables)		N/A	67.47	74.4	72.26
		NB		Macroaveraged		N/A	73.96	66.79	70.19
				Microaveraged		N/A	68.3	66.11	67.19
Troussas et al. [36]	Assign docs sentiments	NB	Facebook 7000 status updates from 90 users	3-fold cross validation	1131 (+)	N/A	77	68	72
					1131 (−)
Singh et al. [37]	Assign docs sentiments	NB	Product review, movie review	4-fold cross validation	N/A	45.6	83.1	N/A	81.2
		J-48				96.7	87.7	N/A	91.7
		BFTree				89.2	88.3	N/A	72.1
		OneR				90	91.3	N/A	97
Rathi [38]	Assign adjectives ±	SVM + KNN, hybrid	Tweet	N/A	267 (+)	76.17	N/A	N/A	N/A
					264 (−)
					168 (0)