Contextual Emotion Classification in Text Using Hybrid Word2Vec-BiLSTM

In the modern digital era, human communication has shifted significantly toward unstructured text found on social media, micro-blogs, and instant messaging platforms. While traditional face-to-face interactions rely on non-verbal cues like facial expressions and vocal tone, digital text lacks these signals, making the detection of emotional intent a complex challenge for Natural Language Processing (NLP). This study addresses the limitations of traditional machine learning methods, such as Naïve Bayes and Logistic Regression, which often fail to capture contextual meaning and long-range sequential dependencies. The proposed research introduces a hybrid deep learning framework that integrates Word2Vec(Continuous Bag-of-Words) embeddings with a Bidirectional Long Short-Term Memory (BiLSTM) network. Word2Vec is utilized to extract dense semantic features, enabling the model to understand mathematical relationships between words.