Table 1 Summary of existing studies
From: Analysis of customer reviews with an improved VADER lexicon classifier
Sr. No | References | Description | Techniques | Dataset | Limitations |
|---|---|---|---|---|---|
1 | Styawati et al. [19] | Sentiment analysis on online transportation service applications reviews of users | Word2vec text embedding model, SVM | Online customer reviews in transportation systems | Grab app gets a lower performance, with 87% accuracy achieved |
2 | Kayıkçı et al. [20] | To understand the public opinion on the recently implemented demonetization policy using the proposed SenDemonNet | F-WOA, HDNN | Twitter sentiment reviews | This work does not consider sarcasm tweets; it considers fewer amounts of tweets |
3 | Nasfi et al. [21] | Sentiment analysis on movie reviews and product reviews | Gaussian-based HMM, SVM | Amazon products and IMDb movie reviews | The hybrid model, especially when using SVMs, may make it susceptible to overfitting |
4 | Sagarino et al. [22] | Sentiment analysis on product reviews as customer recommendations in Shopee Philippines | VADER and MNB | Shopee customer reviews | Generalizability to other e-commerce platforms or languages may be limited, and low accuracy may be achieved |
5 | Benarafa et al. [23] | Sentiment analysis of electronic products and restaurant reviews | Implicit Aspect Identification (IAI) and KNN | Electronic products and restaurant reviews | There is a risk that it might overfit the WordNet data |
6 | Sharma et al. [33] | Domain-specific sentiment analysis using product star ratings | SentiDraw framework | Cornell Movie Reviews Dataset and Large Movie Review Dataset | The average accuracy of the model was below 90% |
7 | Beigi et al. [34] | Domain-specific feature creation and sentiment classification | Domain-Independent Lexicon and Multilayer perception | Amazon multi-domain customer reviews | Each domain utilizes a different range of vocabulary from one another |
8 | Hasanati et al. [35] | Tweets about the COVID vaccination are subject to fine-grained sentiment analysis | SEMMA and SVM | Twitter customer reviews | Not considered domain-specific words relating to COVID-19 and vaccinations |
9 | Juanita et al. [36] | Sentiment analysis on E-Marketplace users' opinions | Lexicon-Based Model and Naive Bayes model | Online customer reviews | Overfitting to the training data |
10 | Thangavel et al. [37] | Lexicon-based procedure to implement tweets sentiment analysis | Lexicon based framework | Twitter customer reviews, STS-Gold Dataset | considers fewer tweets |
11 | Tahayna et al. [38] | To improve the classification of idiomatic tweets with tiny training samples | Lexicon-based approach and Pliable augmentation technique | Twitter customer reviews | 16% error rate to handle larger datasets efficiently |
12 | Ojeda-Hernández et al. [39] | Sentiment analysis to create dictionaries for classification | Formal Concept Analysis (FCA) | Twitter dataset | FCA can be computationally intensive, especially when dealing with large datasets or complex concept lattices |
13 | Yue et al. [40] | Multi-domain sentiment analysis based on neural network | Attention Neural Networks | Amazon and JD customer reviews | computing complexity is higher |
14 | Badr et al.[41] | Labeled sentiment analysis in an unsupervised domain | Unsupervised DomainAdaptation with Source Preservation | Amazon, FDU-MTL, and Spam dataset | Implicit context-specific feature selection can be addressed |
15 | Geethapriya et al. [42] | An improved method in cross-domain sentiment classification | CDSARFE approach | Amazon multi-domain customer reviews | Not considered implicit feature extraction |