Towards a dataset for arabic multimodal sentiment analysis

Abstract

All around the world, Individuals are consistently imparting their insights, stories, and audits through different social media platforms. Concentrating on feeling and subjectivity in these assessment recordings is encountering a developing consideration from the scholarly community and industry. While sentiment analysis has been successful in different languages, it is now an understudied research question for videos and multimedia content in Arabic. The greatest mishaps for concentrating on this path are the absence of a legitimate dataset, technique, baselines, and factual investigation of how data from various methodology sources connect with one another. In summary, this thesis introduces the AMMD dataset, a valuable resource for sentiment analysis in Arabic. It also, encapsulates the essence of the research, highlighting the importance of a multimodal approach and the comprehensive methodology employed for data collection and alignment. The results presented in this thesis offer insights into sentiment analysis in Arabic and pave the way for further advancements in this feld. The AMMD dataset is expected to facilitate research and development of sentiment analysis applications for the Arabic language.