IndexLiterature ReviewConvolutional Neural Networks (CNN)Recursive Neural Network (RNN) and Recurrent Neural Network (Rec NN)Deep Neural Networks (DNN)Deep Neural Networks (DBN)Hybrid Neural NetworksOther Neural networksObservationPositive reviewNegative reviewConclusionThe process of analyzing and classifying words based on the tags associated with the words is called sentiment analysis. As the above statement clearly states that we will use the lexicon based approach i.e. there is a dictionary containing the bag of words (tags) needed for the neural network to analyze whether the sentiment in the text is a positive or negative one. Sentiment analysis has found its applications in many areas of movie review analysis, spam email detection, product success through trend prediction, and so on. The Naïve Bayes classifier is a probabilistic classifier that analyzes the presence of a given sentiment in a sentence. A tokenizer is used to tokenize the sentence and pass the value of the semantic vector to the neural network. The neural network then uses a Naïve Bayes classifier to classify whether the overall sentiment of the sentence is positive or negative. The dataset used is a corpus that contains a lot of labeled words. Say no to plagiarism. Get a tailor-made essay on "Why Violent Video Games Shouldn't Be Banned"? Get an original essay Peer reviewed under the responsibility of the International Conference on Sustainable Computing in Science, Technology and Management. Sentiment analysis research has been going on over the centuries and the main reason for this is that data is readily available in the form of reviews, feedback comments, etc. But this information can be exploited by a machine using a deep learning neural network. A deep learning code combined with a neural network makes the system more dynamic and adaptable. This helps the neural network to analyze various convoluted patterns and also classify them in the presence of noise. A neural network was designed based on the observation of the complexity present in the neurons of biological organisms. This neuron mainly consisted of 3 parts, namely the dendrites, stoma and terminal. There are various types of neural networks available and based on their structure and adaptability they can be used for a myriad of tasks. For example, a CNN is primarily used for pattern detection and classification in an image, while an RNN is used for text due to its exhaustive iterations and high adaptability in the presence of noise. The neural network that we will use in this project will be Naïve Bayes Classifier. The main reason we use Naïve Bayes is that it is designed for sentiment analysis of huge datasets. The naive Bayes classifier classifies each vector tag based on its maximum likelihood. Since it is probabilistic it considers the presence of a feature to be independent of another feature. Maximum likelihood is an estimate of the probability of a particular condition, that is, the probability that a particular event will occur. Since it uses a dictionary which in our case is a corpus, it follows supervised learning. Supervised learning is the process where the neural network learns in the presence of a teacher (image, data, or in our case, tags/labeled data in the bag of words). Based on the semantic vector classified by the neural network, the sentiment is analyzed accordingly. Sentiment analysis has found its application in a variety of applications such as text classification for spam detection,topic detection and recommendations, etc. Furthermore, sentiment analysis when applied with the help of neural network combinations can perform various complicated tasks such as automated responses whose relevance increases with decreasing unavailable time resource. Our proposed model was able to analyze sentiments from the corpus dataset with 73% accuracy. Literature Review Sentiment analysis has received attention from many authors over the past two decades. In recent years, many researchers have contributed several neural network models that constitute the parts of artificial neural networks (ANNs). These models include Convolutional Neural Networks (CNN), Recursive Neural Networks (RNN), Deep Neural Networks (DNN), Recurrent Neural Networks (RNN), and Deep Neural Networks (DBN)[1].Convolutional Neural Networks (CNN)The CNN ( Convolutional Neural Network) [25] includes layers of clustering and sophistication as it provides a standard architecture for mapping variable-length sentences into fixed-size sparse vector sentences. Textual sentiment analysis was performed by the authors [27] on Twitter using deep learning. Work was done to avoid the new feature requirement by initializing the weight of the CNN parameters and it seemed critical to train the model. Neural language and large group of unsupervised tweets were used for word initialization and training, respectively. Sentiment analysis was performed at the message and sentence level and both produced an average score of 1 across all test sets. Sentiment analysis on micro-blogs has been examined in detail by [28]. CNN was used to obtain users' opinions and attitudes regarding hot events. The problem of explicit feature extraction was overcome by using CNN. Implicit learning was achieved by CNN. To collect data from the target, input URL and focused crawler were used, 1,000 to 1,500 blog comments/reviews were collected as a corpus and divided into three labels, namely neutral sentiment, negative sentiment and positive sentiment . The proposed model was compared with previous studies since these studies used additional CRF, SVM, and traditional algorithms to perform sentiment analysis at a high cost. However, the performance shows that the proposed model is reasonable and sufficient to improve the accuracy in terms of sentiment analysis. The combined textual and visual sentiment analysis proposed by researchers in [29] motivated the need to control comprehensive social media content. A convolutional neural network (CNN) is one of the neural networks that works brilliantly when dealing with images. This model was a rule-based sentiment classifier called “VADER”. The tweets were collected via the Twitter API. Sentiment labels for the chosen tweets were constructed using Mechanical Turk (AMT) and crowd intelligence. The results suggest that the joint textual-visual model performed better than the individual visual and textual sentiment analysis models. In the study of [15], researchers depicted a seven-level structure to analyze the sentiments of sentences. This framework depends on CNN and Word2vec for sentiment analysis and computing vector representation, respectively. Dropout, Normalization and Parametric Rectified Linear Unit (PRLU) technology were used to improve the correctness and generalizability of the proposed model. The framework was verified on the rottentomatoes.com dataset which contains the corpus of film review excerpts, thedataset consists of five labels positive, somewhat positive, neutral, negative and somewhat negative. Comparing the proposed PRLU model with previously mentioned models such as Matrix-Vector Recursive Neural Network (MV-RNN) and Recursive Neural Network (RNN) we can observe that the proposed PRLU model outperforms previous Matrix-Vector models with an accuracy of 45 ,5% .Recursive Neural Network (RNN) and Recursive Neural Network (Rec NN) Recursive neural network (RNN) [25] comes under supervised learning. It has a tree structure that settles before training and the nodes have random matrices. Reconstruction of the input in RNN is not necessary. Sentiment Treebank was introduced in the study [31]. Includes fine-grained sentiment labels for sentences in sentence parse trees. To address them, the recursive neural tensor network was introduced. When the proposed model is trained on the new Treebank, this model outperforms all previously mentioned methods. The combination of the new model and data results in a single-sentence sentiment detection system that pushes the state-of-the-art by 5.4% for positive/negative sentence classification. In addition to this standard setting, the dataset also poses important new challenges and enables new evaluation metrics. For example, the RNTN achieves an accuracy of 80.7% on detailed sentiment prediction for all used sentences and also captures the negation of different sentiments. The research of [25], the idea of ​​sentiment analysis with the help of different Recursive architectures: Recurrent neural networks have been proposed. They separated each sentence from each other in the review and handed it to a recursive neural network (RNN). From here the class is decided and the average semantic vector is then analyzed by the neural network to find the sentiment of the statement. This research paper compared various techniques and the conclusion was that Support Vector Machine (SVM) –Linear classifier is more accurate than RNN and RecNN, e.g. Recursive and recurrent neural network architecture respectively. Deep Neural Networks (DNN) In this study [33], the author proposed a model for sentiment analysis considering both visual and textual contents of social networks. This new scheme used a deep neural network model such as Denoising and skip gram autoencoders which is the basic scheme of the Continuous Bag-Of-Words (CBOW) model. The proposed model consisted of two parts CBOW-LR (logistic regression) for textual content analysis which were then expanded to CBOW-DA-LR. This model was able to classify feelings based on the polarity of visual and textual information. Four datasets were evaluated, namely Sanders Corpus, Sentiment140, SemEval2013 and SentiBank Twitter, from which the proposed model outperformed CBOWS+SVM and FSLM (fully supervised probabilistic language model). Perhaps the extended and fully supervised probabilistic language model in terms of small training data had outperformed the current model. As expected, both feature learning and gram skipping required large datasets for optimal performance. Deep Belief Networks (DBNs) Deep belief networks (DBNs) [38] include several hidden layers, composed of RBMs (restricted Boltzmann machines). DBN has been proven to be efficient for feature representation using unlabeled datasets (or raw data) and addresses the shortcomings of labeled analytics problems. In this article [39], a new deep neural network structure called WSDNN (Weakly Shared Deep Neural Networks) was presented. The main goal of WSDNNs is to use two languages ​​to share information.