Sentiment Analysis of Social Media Reviews using QOS Parameterization

The exponential growth of content on social media raised the need for evaluation of user reviews to recognize the underlying sentiments. The traditional Natural Language Processing (NLP) techniques necessitate novel Quality of Service (QoS) parameters from the aspect based reviews. The classical methods espouse QoS parameters acquired from feedback system where, a predefined range of questions affects the authenticity of sentiments. This paper proposes the method of evaluation that assimilates aspect related QoS parameters obtained from user reviews. The pre-processing phase of our proposed model involves steps like review cleaning followed by word tokenization, stemming, and stop-word removal. Pre-processed set of word tokens go through Parts Of Speech (POS) tagging using Stanford POS tagger. Post-processing phase leverages standard NLP and Machine Learning (ML) techniques to identify the prominent QoS features. However, the task of sentiment classification exploits Natural Language Toolkit (NLTK) but, the impact of relevant terms in a review is learned using Logistic Regression (LR). The efficacy of proposed model is evaluated using a real world dataset and the results confirm the effectiveness of introduced QoS features.