Sarcasm detection can help refine traditional sentiment analysis tools by identifying instances where positive or neutral language conveys negative sentiment. This can lead to more accurate insights into public opinion and brand perception on social media. Practical applications of our project include the development of customer service automation, market research, sentiment analysis, and content moderation. The existing system of detecting sarcasm tweets is done by Humans inspection and it has some limit to do the sarcasm tweets and do not reach the milestone to detect the accuracy. Traditional methods lack the ability to discern sarcastic intent, resulting in misclassifications and misinterpretations. To predict the accurate detection and interpretation of sarcasm. It often relies on fine linguistic cues and context, making it difficult for conventional lexical analysis techniques to identify. As a result, there is a need for a more nuanced approach to lexical analysis that can effectively detect and interpret sarcastic expressions. the proposed method, we utilize natural language processing (NLP) techniques for text preprocessing we employ TF-IDF for feature extraction. We utilize a lexical algorithm to detect sarcasm in tweets. Finally, we evaluate the accuracy of sarcasm detection to assess the system's performance. our project aims to introduce the lexical algorithm that addresses the limitations of existing methods and offers improved accuracy and nuanced interpretation of text.