A Hybrid Bag-Boost algorithm for Feature-Based Sentimental Analysis on Mobile Reviews
Siva Kumar Pathuri, Neelamegam Anbazhagan, Mandal K, Thotakura Venkata Sai Krishna, Chitturi Prasad · 2021 2nd International Conference on Smart Electronics and Communication (ICOSEC) · 2021
One of the NLP’s big problems (natural language processing) is a sentimental interpretation or opinion mining. Market analytics is playing a crucial role in the present scenario, perceiving that people are keen to increase their businesses. These people rely in particular on input from products used by the consumers to survive in the market and mining of information gives them an excellent picture of what they can expect in the future. Few words or sentences will choose consequences or effects. So, these people try to improve their business by selling luxury products to provide maximum benefit for their customers. Hence, sentimental analysis in the current years has acquired much attention. SA is an NLP research field used to categorize the perspective or the view inside a text of a particular feature. In addition, the data collection contains a variety of machine learning algorithms and the outcomes are correlated with the Decision Tree classifiers, Naive Bayes which are tested based on parameters such as recall, precision, and F- score. This paper is based on the various classification approaches for deciding whether or not the general sentimentality of a person is undesirable, positive, or impersonal according to the views expressed by the consumers and also predicts the star rating of a mobile. The two advanced methods such as feature classification followed by polarization classification along with the experimental results are also considered. In conclusion, in this paper a comparative analysis is performed among 3 classification techniques 1) Decision Tree, 2) Hybrid-Bag Boost algorithm, 3) Naive-Bayer’s where the hybrid algorithm is of high precision in comparison with the other two algorithms in machine learning. The key goal of the proposed method is to establish a standard for the assessment and the classification based on the analysis text.