Arabic Sentiment Analysis: A Systematic Literature Review

Table 5

Overview of the data extracted from related articles in ASA.

Ref	SA task	Ap	Algorithm	SA level	DS/size	Domain	Language	Features	Accuracy (%)
[17]	AD	Un	Syntax-based approach	Aspect	15000, 15000	Hotel, products	MSA, DA		65.32
[28]	BR&SC	Su	SVM	Aspect	5K	Airline	MSA, DA	Skip-gram, FastText	89
[34]	BR&SC	Su	SVM		363		MSA, DA		76.09
[25]	BR&SC	Su	KNN, NB, SVM		3015	Food, sports, weather	MSA, DA		76.33
[109]	SC	Su	Bagging, NB, KNN, DT		D10, D11, D12		MSA, DA	SMOTE	74.37;74.9, 84.02
[40]	BR&SC	LB	Corpus and LB approach	Word	15,274	13 domains	MSA, DA, CA	TF-IDF, N-gram	93
[43]	BR&SC	Su	SVM	Sentence	D23, D05, D03, D17, D20		MSA, DA	TF-IDF, LSA, CBOW, SG	83.02
[110]	BR&SC	Su	DCNN	Sentence	2390	News	MSA	Word2ve, N-gram	69.9
[44]	BR&SC	Su	SVM, LGR		183531	Products	MSA	BoW, TF-IDF	Bal:77.76; Unbal:91.21
[16]	BR&SC	Su	NB, SVM, RFT		10254	Elections	MSA, DA	N-gram, TF, TF-IDF	77
[49]	BR&SC	Su	SVM, KNN, BNB		5986		MSA, DA	N-gram, TF, TF-IDF	F-m = 88.8
[45]	SC	Su	RFT, GNB, LGR, SGD		D09, D05		MSA, DA	CBOW, SG	87.10
[53]	SC	Hb	Bagging, SVM, RFT, NB		D10, D11		MSA, DA	Unigram, bigram	90.4; 90
[55]	SC	LB	SVM, ME, Bagging, Boosting, RFT, NNET, DT, NB		D16, D03, lex = 5376				95.98
[111]	SC	Su	NB, NB-MLP		2154; 13420; 1353; 3962; 8522	Attraction, hotel, movies, products, restaurants	MSA, DA		99.8; 85.1; 95.4; 97.3; 93.1
[50]	BR&SC	Su	NB, SVM		48	Stock market	MSA, DA	N-gram, TF-IDF, BTO	83.58
[51]	BR&SC	Hb	LB, LSTM, CNN, SVM, LGR, NB, DT, RFT		D24, D05		MSA, DA	TF-IDF, CBOW	87.5; 81
[57]	Subj C	Su	DL		1100		MSA, DA	TF-IDF	92.96
[65]	SC	Su	NB, SVM		D09	Books	MSA, DA	TF-IDF, N-gram	90.98
[67]	BR&SC	Su	NB, SVM			News for multidomains	MSA, DA	Unigram, bigram, TF, TF-IDF	84,56
[112]	SC	Su	CNN-LSTM	Character, Ch5Gram, Word	D01, D03, D05	Health	MSA	Trigram, ReLU	95.68; 77.62; 94.24; 88.10
[69]	BR&SC	Hb	SO-NB	Sentence	1200		MSA	N-gram models, semantic features	90
[79]	BR&SC	Hb	LGR, PAG, SVM, PRN, RFT, ABT, LB	Review	D13	Hotel	MSA, DA	Unigram, bigram, semantic features, BOW	94 to 97
[113]	Aspect SA	Su	NB, BYN, DT, KNN, SVM	Sentence	D19	Hotel	MSA	Morphological, syntactic, semantic features	95.4
[56]	BR&SC	Su	NB, SVM, DT, RFT		1543347; 1462	Politics	MSA	Syntactic, surface-form, sentiment features	71.95
[47]	BR&SC	Su	NB	Sentence	18278	Elections	MSA, DA	Unigram, bigram, IG	93.13
[59]	BR&SC	Su	SVM, NB	Document	D21	News, politics, sports, culture	MSA, DA	Unigram, bigram	90.20
[46]	BR&SC	Hb	SO, SVM, NB	Sentence	1520		DA	N-gram, CountVectorizer	92.98
[114]	SC	Su	CNN	Sentence	D05, D20		MSA, DA	Glove, SG, CBOW, random word vectors	72.1463
[60]	BR&SC	Su	SVM, NB, KNN	Sentence	996	Multidomain	MSA, DA	Unigram, bigram, BTO, TF-IDF	78
[73]	SC	Su	DT	Document	D16, D06	Movies, hotel	MSA, DA		93.83; 90.63
[80]	Subj C, &SC	Hb	DT, NB, KNN, Ontology Baseline, LB	Aspect	D09, 2000	Books, Hotel	MSA	Domain features	f-m = 79.18;78.83
[81]	BR&SC	Hb	SVM, NB, KNN	Sentence	3476	Movies, economy, sports, history, politics	MSA, DA		97.44
[61]	BR&SC	Su	SVM, KNN, NB, DT, LEM2	Document	4812		DA	Rough set method	74
[66]	BR&SC	Su	RNTN	Phrase, sentence	1177		MSA	Orthographic, morphological features	80
[35]	SC	Su	SVM, NB, MNB, SGD, DT	Review	D09	Books	MSA, DA	Unigram, bigram, TF-IDF, genetic algorithm	94
[32]	BR&SC	Un	LB	Document	1000, 1000	News, arts	DA		73–96
[15]	BR&SC	Su	SVM	Sentence, document	D19	Hotel	MSA	Unigram	76.42
[83]	BR&SC	Hb	KNN, SVM, LLR, NB, NEUNET	Document	8861	15 domains	DA	Polarity scores, dialects, synsets, inflected forms	97.8
[84]	Aspect SA	Un	LB	Review	200		MSA	Sentiment features	92.15
[86]	BR&SC	Hb	SVM, NB, LB	Sentence	64342		DA	N-gram	70
[115]	SC	Su	Stacking, SGD, RFT, LGR, GNB	Sentence	1350		DA	Word2vec, SMOTE	85.28
[54]	SC	Su	SVM, NB	Sentence	9096		DA	BOW, bigram, trigram	f-m = 73
[68]	BR&SC	Su	SVM, NB	Sentence	1800		MSA, DA	N-gram models, TF-IDF	88.72
[88]	SC	Su	MLP, LGR	Sentence	D09		MSA, DA	Doc2vec	32.38
[90]	SC	Su	CNN, LSTM, RNN	Document	D05, D03		DA	CBOW, SG, mul, CONC	81.63; 87.27
[82]	SC	Su	SVM, DT, NB, KNN, HCs	Document	D09		MSA, DA	BOW, correlation analysis	72.64
[48]	BR&SC	Su	DT, RFT, SVM	Sentence	10254	Politics	MSA, DA	N-gram models, TF-IDF, TF	81
[91]	SC	Su	DNN (LSTM), RNN	Sentence	D09		MSA, DA	Word embedding, BOW	71
[14]	SC	Su	RNN, SVM	Sentence	D19		MSA	word2vec, lexical, morphological, semantic features, N-gram	95.4
[116]	Subj C&SC	Su	SVM	Sentence	D04		MSA		69.37
[104]	BR&SC	Su	SVM, NB	Sentence	1121	Education	MSA	N-gram	84.62
[71]	SC	Su	SVM, NB	Review	D16	Movies	MSA	N-gram models, TF-IDF, BTO, TF	96.67
[29]	SC	Su	SVM, LGD	Review	Unbal, bal (D09)	Books	MSA, DA	BOW, N-gram,	88.51, 78.14
[30]	BR&SC	Su	LGR, PAG, SVM, PRN	Review	Unbal, bale (D07)	Books	MSA, DA	Unigram, bigram	0.744–0.911; 0.847–0.85
[13]	BR&SC	Su	SVM, NB	Document	1331	Products	MSA, DA	N-gram, BTO; TF-IDF	89.68
[89]	SC	Su	SVM.NB, KNN, ANN	Sentence	500	Hotel	MSA		f-M = 92
[36]	BR&SC, Subj C	Su	SVM, NB	Sentence	3700		MSA, DA	N-gram models	89.55
[58]	SC	Su	AR		D16	Movies	MSA, DA	IG, chi-square, GI	86.81
[93]	BR&SC	Su	Lexical semantic, CRF	Sentence	381	News	MSA	Position, bigram, trigram, morphological	f-m = 84.93
[94]	BR&SC	Su	SVM, NB, KNN	Document	250	Products	MSA, DA	N-gram models	94
[95]	BR&SC	Su	SVM, MNB	Sentence	134194		MSA, DA	Automatic labeling	75.7
[117]	SC	Su	Rule-based approach	Sentence	D16, 2000	Movies, arts, politics	MSA, DA		85.6, 93.9
[97]	BR&SC, Subj C	Su	SVM, NB, DT, RFT	Sentence		Politics	MSA	N-gram, TF-IDF	P = 70.97
[98]	SC	Su	SVM, KNN, NB, DT, RFT	Sentence		Terrorism	MSA	Lexical, surface-form, N-gram	P = 71.76
[99]	SC	Su	SVM	Aspect	Restaurant reviews, 1000, D16,	Novels, products, movies, sports, hotels, restaurants	MSA	N-gram	85.35; 90.60; 96.00
[100]	BR&SC	Su	SVM, BPNN, NB, DT	Sentence	2000		MSA, DA		96.06
[62]	BR&SC	Su	SVM, NB	Sentence	18278	Politics	DA	N-gram, TF-IDF
[118]	SC	Su	SVM, CNB, MNB	Sentence	D10, D11, D12		MSA	N-gram, TF, IDF, TF-IDF, IG	77.34
[37]	BR&SC	Hb	SO-SVM-NB	Sentence	4800	News	DA	N-gram models	80.9
[92]	SC	Su	SVM, NB, KNN	Sentence	3073	Politics, arts	MSA, DA	N-gram	68.69
[74]	BR&SC	Su	SVM, NB, KNN	Sentence	2,591	Education, sports, politics	MSA	BTO, TF-IDF, TF	69.97
[63]	BR&SC	LB	LB	Sentence, document	D09, D14	Health, book	MSA, DA		71
[38]	BR&SC	Su	NB, SVM, ME	Document	28576		DA	BoW, N-gram models	86.75
[41]	SC	Su	SVM, NB, DT	Sentence	2000	Politics, arts	MSA		P = 80.9
[101]	BR&SC	Hb	LB, NB	Sentence	2590	Restaurants	DA	Sentiment words	90.54
[102]	BR&SC	Su	NB, DT	Sentence	2000		MSA		64.85
[33]	BR,&SC, Subj C	Su	RFT, GNB, SVM, LGR, SGD	Word	D05, D03, D17, D09	Quran, customers	MSA, DA	CBOW	80.21–81.69
[119]	SC	LB	Rule-based, LB		D16		MSA		89.6
[18]	BR&SC	Su	CRF, DT, NB, KNN	Sentence	2265	News	MSA	NER, N-gram	86.5
[42]	BR&SC	Hb	SO, SVM	Sentence	1103		MSA, DA	N-gram models	84.01
[103]	BR&SC	LB	LB	Word	D25, D05, D18, Lex: D25-PMI		DA		89.58
[87]	BR&SC	Su	SVM	Sentence	625	Multidomain	MSA, DA		83.5
[96]	BR&SC	Su	Bagging, boosting	Sentence	1500	Sports, news, Economics	MSA, DA		85.95
[72]	BR&SC	Su	SVM, NB	Document	D02		MSA, DA		M-P = 90.5
[31]	BR&SC	Su	SVM-KNN	Review	625	Hotel	DA		97
[105]	SC	Su	DT, Dtable, SVM, MNB, voting (KNN, DT, and NB)	Sentence	Unbal, bal : D09		MSA	BOW	42.7–46.4
[120]	BR,&SC, Subj C	Su	SVM, MNB, BNB, PAG, SGD, LGR, PRN, KNN	Sentence	D05		DA	N-gram models	69.1
[121]	BR&SC	Su	Voting (RSS and SVM)	Sentence	800	News	MSA, DA		98
[64]	SC	Su	Voting (ME, SVM, and ANN), bagging, boosting, stacking	Document	D16, D11-D13		MSA	Stylistic, morpholexical, tigram, brigram, TF-IDF, TF, BTO	F1 = 85.06
[122]	BR&SC	Ss	SVM, LGR, BNB, KNN, SGD	Document	D09	Movies, hotels, restaurants, products	MSA	Unigram, bigram, TF-IDF, word count	82.4
[106]	BR&SC	Su	SVM, NB, KNN	Sentence	500		MSA		f-m = 91.5
[70]	BR&SC	Su	NB, KNN	Document	2591	Education, sports, politics	MSA, DA	Bigram	M-P = 75.25
[26]	BR&SC	Su	SVM, NB, KNN	Sentence	D16, 164	Politics	MSA, DA	Correlation, N-gram models	96.6
[107]	BR&SC	Su	NB, KNN	Sentence	300		MSA		63.79
[75]	BR&SC	Su	SVM, NB	Sentence	300, 250		DA	N-gram models	0.75.64
[123]	SC	Su	SVM, NB	Document	5070	News, entertainment, sports, science, business	MSA	Chi-square, correlation, GSS coefficient, IG, relief F	M-F = 95.1
[76]	BR&SC	Su	NB, voting	Document	4812		DA	Rough set method, genetic	57
[52]	BR&SC	Su	SVM, NB	Document	D22, D16, 7400		MSA	BTO, TF, TF-IDF, score	94.88–97.81
[124]	BR&SC	Ss	Pattern matching, majority with entities	Sentence	5000	Restaurants	DA		60.5
[24]	Subj C, and SC	Su	SVM, NB		Obama, Messi, iPhone, shia	Politics, social, products	MSA	N-gram, IDF, TF-IDF	68.05; 87.43
[85]	BR&SC	Su	SVM, MNB	Sentence	260		DA	N-gram	72.78
[39]	BR&SC	Su	SVM	Sentence	1350	News	DA		P = 88.63
[77]	SC	Hb	Voting (SVM, NB, and SO)	Sentence	D16, D22		MSA	N-gram models	F1 = 90.74
[78]	SC	Su	NB, SVM, KNN	Document	D16, D10, D11		MSA	TF-IDF, TF, IDF, N-gram	93.0, 78.2
[125]	SC	Su	MNB, BNB, SVM	Sentence	Unbal, bal : D09		MSA, DA		91.0, 82.7
[27]	BR&SC	Su	SVM, NB	Sentence	Unbal, bal: 4625	Arts, politics, science, technology, social	MSA, DA		68.2, 61.4
[12]	BR&SC	Su	KNN, NB, DT, SVM	Sentence	1000		MSA	Social mention, Senti. strength	95.59, 93.29