Sentiment analysis using hybrid Deep learning approaches

Abstract

NLP (Natural Language Processing) has made significant strides during the past ten years. Sentiment analysis is among the most active and popular NLP research fields. In the research field of natural language processing of textual data, sentiment analysis and its application for business analytics have drawn increasing interest from organizations, enterprises, and researchers in recent years. This proliferation is due to the fact that the sentiments expressed play key role on almost every human endeavor and a considerable influence on our behavior. Text sentiment analysis demands the use of sophisticated natural language processing methods in conjunction with state-of-the-art machine learning algorithms that can learn from both structured and unstructured data. As a result, different natural language processing and machine learning based algorithms have been employed in the past to analyze sentiments at the document, phrase, and aspect level. The primary drawback of the lexicon-based methodology is the inaccurate sentiment scoring of opinion words by the current lexicons, such as SentiWordNet whereas machine learning methods need domain expertise, the extraction and presentation of the handcrafted feature engineering. However, deep learning-based approaches are gaining a lot of popularity because of their excellent performance in recent years in many application domains, from computer vision, speech recognition to NLP. Numerous text mining issues, such as clustering, document classification, web mining, document summarization, and sentiment analysis, have found successful solutions due to deep learning approaches. Each deep learning method has unique benefits and drawbacks. A single deep learning method is not enough in order to gain better performance result when used in certain areas. As a result, researchers have focused on hybrid deep learning models in order to gain advantage of the models involved in the hybridization. To overcome the problem of the traditional lexicon-based approach and machine