Download PDFOpen PDF in browserA Review of Opinion Mining TechniquesEasyChair Preprint 68719 pages•Date: October 19, 2021AbstractOpinion mining or Sentimental analysis in one of the recent challenge in Natural Language Processing (NLP). Individual express their opinion in different platforms like Facebook, Twitter, Yelp is also a challenging task as the innovation which has been increased in an exponential way. With the growth of social media, a large amount of data namely as comments, reviews and opinions have been generated. According to researchers, analysis of sentiments is based on sentence level, document level, aspect level and user level. The analysis of this data consumes more time and it is difficult for processing. So, there is a need to develop an intelligent system that will either classify or determine positive, negative, and neutral opinions. The main aim of this paper is to review the opinion mining concepts with techniques namely as machine learning, deep learning, transfer learning and Hadoop framework briefly. With standard machine learning algorithms, the most time-consuming process are feature engineering and feature extraction. As a result, deep learning alleviates the burden of feature creation because as the network learns, it produces the features on its own. The lack of labelled data makes it difficult to train NLP models. Transfer learning is one of the most effective solutions to this problem. There is no need to start from scratch when training AI models. Transfer learning has several advantages, including shorter training times, higher output accuracy, and the need for less Sentiment analysis can be carried out without sacrificing accuracy or speed. It can scale to larger data sets while maintaining performance. Hadoop-implemented opinion mining in less complex, more easily extendable, and high performance at a lower cost. Keyphrases: Hadoop, NLP, Opinion Mining, Transfer Learning, deep learning, machine learning
|