Deep learning based sentiment analysis for cloud provider selection

The selection of a viable Cloud Service Provider (CSP) has always been a crucial task for a Cloud Service Consumer (CSC) to avail of their offered services. This selection would enable a service consumer to maintain a trustful relationship with a provider. For that purpose, consumer reviews posted on internet websites and other social media platforms need to be carefully evaluated for a proper CSP selection. Sentiment Analysis, also termed Opinion Mining, is the computational treatment of text's views, experiences, sentiments, and subjectivity. Aspect-Based Sentiment Analysis (ABSA) extracts informative aspects within the text and uses them to classify the sentiment of reviews. Nowadays, different lexicon- based, supervised learning, and un-supervised learning techniques are used for sentiment classification tasks. Deep Learning is an AI technique used for language processing, text analysis, pattern recognition, sequence predicition tasks, etc. Its types, such as Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), use different strategies to carry out processings. The dissertation performs Aspect-Based Sentiment Analysis of cloud consumer reviews using Deep Learning approaches of RNN, LSTM and GRU. The cloud reviews are extracted using Harvesting-as-a-Service (HaaS) framework. Analytic Hierarchy Process (AHP) model is used to decide the importance and priorities of aspects for Cloud Service Consumers (CSCs). The evaluation would assist cloud service consumers (CSCs) choose the best CSP ideal for their requirements.