Emotions are regarded as being of the utmost importance since they play a significant role in human connection. In today's environment, social media is crucial to how individuals engage with one another. Emotional analysis of these social media posts is beneficial. Twitter is a microblogging platform where users from all over the world may post and express their emotions. Because tweets are brief and informal, sentiment analysis for Twitter communications (also known as "tweets") is thought to be a difficult topic. Recurrent neural networks are used to build and train a model that can identify emotions in tweets. There are thousands of tweets in the dataset, and each one is categorized into one of six emotions: love, fear, joy, sadness, surprise, and rage.
Sentiment analysis, commonly referred to as opinion mining or emotion AI, is the systematic identification, extraction, quantification, and study of affective states and subjective data using natural language processing, text analysis, computational linguistics, and biometrics. Sentiment analysis is frequently used in marketing, customer service, and clinical medical applications. It is used to voice of the customer materials including reviews and survey replies, internet and social media, and healthcare materials. The practice of gathering data about a consumer's assessment of a good, service, or brand is known as sentiment analysis. Natural language processing (NLP) is used in social media sentiment analysis to analyze online mentions and ascertain the emotions that underlie the post. Using social sentiment analysis, you may determine whether a post was favorable, unfavorable, or neutral.
Classifying the polarity of a given text at the document, sentence, or feature/aspect level—whether the conveyed opinion in a document, sentence, or entity feature/aspect is positive, negative, or neutral—is a fundamental task in sentiment analysis. For example, advanced, "beyond polarity" sentiment classification examines emotional states including pleasure, rage, disgust, sadness, fear, and surprise.
In this project, an LSTM-based model is built and trained to identify emotions in tweets. There are thousands of tweets in the dataset, and each one is categorized into one of six emotions: love, fear, joy, sadness, surprise, and rage. This multiclass classification problem of the natural language processing domain is resolved using TensorFlow as the machine learning framework.
Confusion Matrix :
The model provides 99.12% accuracy in training and 89.30% accuracy in Validation