Sentiment Analysis (SA) is a computational study of the sentiments expressed in text toward entit... more Sentiment Analysis (SA) is a computational study of the sentiments expressed in text toward entities (such as news, products, services, organizations, events, etc.) using NLP tools. The conveyed sentiments can be quantified using a simple positive/negative model. A more fine-grained approach known as Multi-Way SA (MWSA) uses a ranking system like the 5-star ranking system. Since in such systems, rankings close to each other can be confusing, it has been suggested that Hierarchical Classifiers (HCs) can outperform traditional Flat Classifier (FCs) for MWSA. Unlike FCs which try to address the entire classification problem at once, HCs utilizes tree structures where the nodes are simple classifiers customized to address a subset of the classification problem. This study aims to explore extensively the use of HCs to address MWSA by studying six different hierarchies. We compare these hierarchies with four well-known classifiers (SVM, Decision Tree, Naive Bayes, and KNN) using many measures such as Precision, Recall, F1, Accuracy and Mean Square Error (MSE). The experiments are conducted on the LABR dataset consisting of 63K book reviews in Arabic. The results show that using some of the proposed HCs yield a significant improvement in accuracy. Specifically, while the best Accuracy and MSE for FC are 45.77% and 1.61, respectively, the best accuracy and MSE for an HC are 72.64% and 0.53, respectively. Also, the results show that, in general, KNN benefitted the most from using hierarchical classification.
Sentiment Analysis (SA) is a computational study of the sentiments expressed in text toward entit... more Sentiment Analysis (SA) is a computational study of the sentiments expressed in text toward entities (such as news, products, services, organizations, events, etc.) using NLP tools. The conveyed sentiments can be quantified using a simple positive/negative model. A more fine-grained approach known as Multi-Way SA (MWSA) uses a ranking system like the 5-star ranking system. Since in such systems, rankings close to each other can be confusing, it has been suggested that Hierarchical Classifiers (HCs) can outperform traditional Flat Classifier (FCs) for MWSA. Unlike FCs which try to address the entire classification problem at once, HCs utilizes tree structures where the nodes are simple classifiers customized to address a subset of the classification problem. This study aims to explore extensively the use of HCs to address MWSA by studying six different hierarchies. We compare these hierarchies with four well-known classifiers (SVM, Decision Tree, Naive Bayes, and KNN) using many measures such as Precision, Recall, F1, Accuracy and Mean Square Error (MSE). The experiments are conducted on the LABR dataset consisting of 63K book reviews in Arabic. The results show that using some of the proposed HCs yield a significant improvement in accuracy. Specifically, while the best Accuracy and MSE for FC are 45.77% and 1.61, respectively, the best accuracy and MSE for an HC are 72.64% and 0.53, respectively. Also, the results show that, in general, KNN benefitted the most from using hierarchical classification.
Uploads
Papers by Mohammed Al-Kabi