Detection of semantic obsessive text in multimedia using machine and deep learning techniques and algorithms

Word boycott has been seen frequently trending in India on various social media platforms. We studied the obsession of Indians with the word boycott; to show the protest or dissent against any government policy, Netflix series, political or religious commentary, and on various other matters, people in India prefer to trend word "Boycott" on multiple mediums. We studied how ingrained the word "Boycott" is in Indians in our research and how it affects daily life, unemployment, and the economy. The data was collected using Youtube API with the next page token to get all the search results. We preprocessed the raw data using different preprocessing methods, which are discussed in the paper. To check our data's consistency, we fed the data into various machine learning algorithms and calculated multiple parameters like accuracy, recall, f1-score. Random forest showed the best accuracy of 90 percent, followed by SVM and Knn algorithms with 88 percent each. We used word cloud to get the most dominant used words, Textblob, for sentiment analysis, which showed the mean Polarity of 0.07777707038498406 and mean subjectivity 0.2588880457405638. We calculated perplexity and coherence score using the LDA model with results -12.569424703238145 and 0.43619951201483725, respectively. This research has observed that the word boycott is a favorite to the Indians who are often using it to show opposition or support related day-to-day matters.

@article{machine2021detection,
  title={Detection of semantic obsessive text in multimedia using machine and deep learning techniques and algorithms},
  author={MACHINE, MULTIMEDIA USING},
  journal={Journal of Theoretical and Applied Information Technology},
  volume={99},
  number={11},
  year={2021}
}