With the rapid advancement of the World Wide Web (WWW), the emergence of social media platforms has increased popularity. People often rely on the web to find information from a diverse range of information sources. The social network comprises an increased amount of opinionated natural language texts, which leverages the different research areas in developing both the individual and business satisfaction aspects. In social media mining, opinion mining has become one of the significant sub-areas of natural language processing. Nowadays, sarcasm is an unconventional way of communication, representing the context with the utilization of the opposite words of the literal meaning. Identifying sarcasm is a difficult task due to the lack of analyzing the gap between the intentional and literal meaning. The social network provides an ideal platform for Internet users to share and exchange information; however, it leads to the spreading of rumors. The existing rumor detection models have analyzed the user profiles, user-generated content, and propagation structure. In addition, mitigating the rumor spread and its effects is critical over the increased dynamics of the rumors and the fast-spreading among the millions of social users. Hence, early identifying the level of rumor truthfulness and preventing its consequences before wide-spreads on social media is essential. To effectively extract the original context of the statement for the different natural language processing-based applications, several researchers have focused on sarcasm detection using different data mining, knowledge-based, machine learning, and deep learning approaches.