Social media platforms like X (formerly Twitter) generate vast volumes of user-generated content that provide real-time insights into public sentiment. Despite the widespread use of traditional machine learning methods, their limitations in capturing contextual nuances in noisy social media text remain a challenge. This study leverages the Sentiment140 dataset, comprising 1.6 million labeled tweets, and develops predictive models for binary sentiment classification using Naive Bayes, Logistic Regression, and the transformer-based BERT model. Experiments were conducted on a balanced subset of 12,000 tweets after comprehensive NLP preprocessing. Evaluation using accuracy, F1-score, and confusion matrices revealed that BERT significantly outperforms traditional models, achieving an accuracy of 89.5% and an F1-score of 0.89 by effectively modeling contextual and semantic nuances. In contrast, Naive Bayes and Logistic Regression demonstrated reasonable but consistently lower performance. To support practical deployment, we introduce SentiFeel, an interactive tool enabling real-time sentiment analysis. While resource constraints limited the dataset size and training epochs, future work will explore full corpus utilization and the inclusion of neutral sentiment classes. These findings underscore the potential of transformer models for enhanced public opinion monitoring, marketing analytics, and policy forecasting.

Sentiment analysis, a vital aspect of natural language processing, involves the application of machine learning models to discern the emotional tone conveyed in textual data. The use case for this type of problem is where businesses can make informed decisions based on customer feedback, identify the sentiments of their employees, and make decisions on hiring or retention, or for that matter, classify a text based on its topic like whether it is about a particular subject like physics or chemistry as is useful in search engines. The model leverages a sequential architecture, transforms words into dense vectors using an Embedding layer, and captures intricate sequential patterns with two Long Short-Term Memory (LSTM) layers. This model aims to effectively classify sentiments in text data using a 50-dimensional embedding dimension and 20 % dropout layers. The use of rectified linear unit (ReLU) activations enhances non-linearity, while the SoftMax activation in the output layer aligns with the multi-class nature of sentiment analysis. Both training and test accuracy were well over 80%.

The task of recommending products to customers based on their interests is important in business. It is possible to accomplish this with machine learning. To reduce human effort by proposing movies based on the user's interests efficiently and effectively without wasting much time in pointless browsing, the movie recommendation system is designed to assist movie aficionados. This work focuses on developing a movie recommender system using a model that incorporates both cosine similarity and sentiment analysis. Cosine similarity is a standard used to determine how similar two items are to one another. An examination of the emotions expressed in a movie review can determine how excellent or negative a review is and, consequently the overall rating for a film. As a result, determining whether a review is favorable or adverse may be automated because the machine learns by training and evaluating the data. Comparing different systems based on content-based approaches will produce results that are increasingly explicit as time passes.

The field of sentiment analysis is a crucial aspect of natural language processing (NPL) and is essential in discovering the emotional undertones within the text data and, hence, capturing public sentiments over a variety of issues. In this regard, this study suggests a deep learning technique for sentiment categorization on a Twitter dataset that is based on Long Short-Term Memory (LSTM) networks. Preprocessing is done comprehensively, feature extraction is done through a bag of words method, and 80-20 data is split using training and testing. The experimental findings demonstrate that the LSTM model outperforms the conventional models, such as SVM and Naïve Bayes, with an F1-score of 99.46%, accuracy of 99.13%, precision of 99.45%, and recall of 99.25%. Additionally, AUC-ROC and PR curves validate the model’s effectiveness. Although, it performs well the model consumes heavy computational resources and longer training time. In summary, the results show that deep learning performs well in sentiment analysis and can be used to social media monitoring, customer feedback evaluation, market sentiment analysis, etc.