Exploring Diverse Conventional and Deep Linguistic Features for Sentiment Analysis of Online Content

Social media has changed the world by providing the facility to common person to share their views and generate their own content, known as Users Generated Content (UGC). Due to huge volume of UGC data being created at great velocity, so to analysis this big data, latest AI (Artificial Intelligence) and its sub-domain NLP (Natural Language Processing) are being used. Sentiment analysis of online content is an active research area due to its vast applications in business for review analysis, social and political issues. In this research study, we aim to carry out sentiment analysis of online content by exploring conventional features like Term Frequency – Inverse Document Frequency (TF-IDF), Count-Vectorization, and state of the art word embeddings based word2vec. Extensive exploratory data analysis has been carried out using the latest data visualization approaches. The main novelty lies in the application of unique and diverse machine learning algorithms on social media datasets and the results evaluation using standard performance evaluation measures reveal that the word2vec using Quadratic Discriminant analysis-based classifier show optimal results.

@article{Tang2025,
  title     = {Exploring Diverse Conventional and Deep Linguistic Features for Sentiment Analysis of Online Content},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {16},
  number    = {1},
  year      = {2025},
  publisher = {The Science and Information Organization},
  author    = {Yajun Tang},
  doi       = {10.14569/IJACSA.2025.0160115},
  url       = {https://doi.org/10.14569/IJACSA.2025.0160115}
}