Propaganda, a form of communication aimed at influencing the opinions and emotions of a target audience, has been a pervasive issue in modern society. With the rapid growth of digital media, the spread of propaganda has become more widespread and difficult to detect. Automated propaganda detection has emerged as an important task in natural language processing (NLP) to help identify and combat the spread of manipulative and misleading information. In this report, we focus on two specific tasks related to propaganda detection: (1) determining whether a given sentence contains propaganda, and (2) identifying the specific propaganda technique used in a span of text known to contain propaganda. We use the dataset provided, which consists of sentences labeled with one of eight propaganda techniques or as "not propaganda." The main objectives of this report are to:

• Develop and evaluate models for detecting the presence of propaganda in sentences. • Build and assess models for classifying specific propaganda techniques in text spans. • Compare the effectiveness of different approaches for each task. • Analyse the errors made by the models and identify potential areas for improvement

Tools/Tech: Python, Numpy, Pandas, Scipy, Nltk, GPU, Transfomer models, Jupyter notebook.