Abstract:Social media is nowadays a vital platform where people can share their feelings about any incident, product, or any issue. Twitter is one of those platforms which are very popular. If we must make use of this to extract some information and to use that to make recommendation systems, to know the mental status of any person or to find the review, we have to perform the sentiment analysis of those tweets. Recently, deep learning approaches emerged as powerful computational models that discover intricate semantic representations of texts automatically from data without feature engineering. This present work discusses how to find the sentiments from a stream of tweets from the twitter, blogging site subjected to pre-processing and classification supported on emotion of the subject matter as "POSITIVE" or "NEGATIVE". We have used a twitter dataset of 1.6 million data consisting of general tweets on various domains with 800,000 POSITIVE and 800,000 NEGATIVE sentiments. We have developed 3 models with Recurrent Neural Network (RNN), Bi-directional RNN, Long Short-Term Memory (LSTM). Our work focuses on the way which will help us to understand the basic sentiment of people which they want to express by their tweets, comment, or blogs. We have obtained highest accuracy of 83.50%, F1 score of 84.60% from our model with LSTM architecture in analysing the sentiment. LSTMs are specifically designed to address the vanishing gradient problem that occurs in traditional RNNs when learning long-term dependencies. In sentiment analysis, understanding the sentiment of a tweet may require capturing dependencies between words that are spread across a potentially long sequence. LSTMs are better at retaining and utilizing information over extended sequences.

Twitter part-of-speech tagging using pre-classification Hidden Markov model

Experimental Study of Hidden Markov Model Based Part-of-speech Tagging for Chinese Texts

Hidden Markov Model Based Part of Speech Tagger for Sinhala Language

Part-of-Speech Tagging with Bidirectional Long Short-Term Memory Recurrent Neural Network

Unknown Word Processing In Hmm-Based Pos Tagging

Sentiment analysis incorporating convolutional neural network into hidden Markov model

A lexicalized second-order-HMM for ambiguity resolution in Chinese segmentation and POS tagging

Massively Parallel Part of Speech Tagging Using Min-Max Modular Neural Networks

TWO-STAGE MODEL SELECTION WITH PARAMETERS WEIGHTED HIDDEN MARKOV MODELS AND LIKELIHOOD RATIO FOR PART-OF-SPEECH TAGGING

Transferring from Formal Newswire Domain with Hypernet for Twitter POS Tagging.

Sentiment Commonsense Induced Sequential Neural Networks for Sentiment Classification.

A Self-Adaptive Hidden Markov Model for Emotion Classification in Chinese Microblogs

Hidden Markov Chains, Entropic Forward-Backward, and Part-Of-Speech Tagging

Token Identification Using HMM and PPM Models

Efficient Part-of-Speech Tagging with a Min-Max Modular Neural-Network Model

Interactive POS-aware network for aspect-level sentiment classification

Hybrid Deep Learning Approach for Sentiment Analysis on Twitter Data

Modeling Sentiment and Aspect Using Syntax: A Topic Model Approach

User-level Sentiment Analysis Incorporating Social Networks

Sentiment Analysis of Microblog text based on joint sentiment-topic model

The Study of a Nonstationary Maximum Entropy Markov Model and Its Application on the Pos-Tagging Task