In today's society, most of the news consumption by people is through different social media platforms, since it is the most easy and convenient way of sharing news to each other. But this becomes the risk in widespread dissemination of fake news. These fake news not just adversely affect an individual but it also affects the society as a whole. Organizations from all sectors are currently having difficulty in finding practical solutions for identifying online fake news, which is a major problem. Subsequently, the false material is frequently posted online to fool users, it is fairly challenging to identify it. Deep Learning (DL) based algorithms can identify fake news more precisely when compared to various Machine Learning (ML) techniques. The current study, the proposed novel hybrid DL model which planning incorporates various word embedding with several DL methods like Recurrent Neural Network (RNN) model such as Long Short Term Memory (LSTM) and Gated Recurrent Unit (GRU). The suggested technique gathers characteristics from several DL word embedding techniques, integrates these features, and categorizes texts according to their Natural Language Processing (NLP) as sentiment polarity. Therefore, it is quickly as possible in recognizing and terminating the negative impacts of false information using Sentimental Analysis (SA). The resulting experiment demonstrated that DL models performed better accuracy than ML classifiers whereas, the accuracy of the LSTM and GRU models is nearly equivalent. The demonstrated possibility to recognize fake news with significant precision by integrating an enriched linguistic set of characteristics with ML and DL models.