A HYBRID DNN-HHO APPROACH FOR EVENT DETECTION IN BIG DATA
K. Swapnika, D. Vasumathi · Indian Journal of Computer Science and Engineering · 2022
Social media are digitally mediated technologies which are interactive, and enable people to build and share content, career interests, ideas, and other modes of expression through virtual networks.Twitter has gained popularity as a medium of social media in recent years.Twitter is being used by users to post on real-life activities.The prime objective of this paper is to use big data to detect certain events in Twitter.Furthermore, considering the large number of tweets, the event detection algorithm must be scalable.This paper attempts to tackle these challenges with the help of DNN and Harris Hawk optimization algorithm.The main steps include in the proposed methodology are, preprocessing of input data, extraction of useful features, feature selection, and classification.Here, the proposed methodology is implemented using PYTHON platform.The performance of the proposed method is analyzed in terms of statistical measures such as F1-score, precision, accuracy, and recall.The proposed method is utilized for detecting five different events such as education, transportation, environment, geospatial and water and the accuracy obtained for each event is 94%, 92%, 98%, 96% and 96.5% respectively.The overall result shows that, our proposed methodology gives better performance in event detection.