Research Article
Temporal Sentiment Analysis (TSMFPMSM) Model for Multimodal Social Media Fake Profile Detection
@INPROCEEDINGS{10.1007/978-3-031-66044-3_25, author={Bhrugumalla L. V. S. Aditya and Sachi Nandan Mohanty and Yalamanchili Salini}, title={Temporal Sentiment Analysis (TSMFPMSM) Model for Multimodal Social Media Fake Profile Detection}, proceedings={Pervasive Knowledge and Collective Intelligence on Web and Social Media. Second EAI International Conference, PerSOM 2023, Hyderabad, India, November 24--25, 2023, Proceedings}, proceedings_a={PERSOM}, year={2024}, month={8}, keywords={social media Temporal Afinn 1D CNN GloVE Word2Vec Interfaces Sentiments}, doi={10.1007/978-3-031-66044-3_25} }- Bhrugumalla L. V. S. Aditya
Sachi Nandan Mohanty
Yalamanchili Salini
Year: 2024
Temporal Sentiment Analysis (TSMFPMSM) Model for Multimodal Social Media Fake Profile Detection
PERSOM
Springer
DOI: 10.1007/978-3-031-66044-3_25
Abstract
Today’s social media sites should be able to spot fake profiles. Most social media accounts (around 25%) are fake or managed by automated software. Therefore, advanced models are required to detect and remove these fake profiles. Anomalies in login, usage, and non-functional elements have inspired researchers to construct pattern analysis models. This book expands on existing models using temporal sentiment analysis to spot fake profiles in social networks with multiple interface types. Massive datasets are gathered from social media platforms like Twitter, Facebook, and others and used in the model. TSPs are derived from these data sets using a unique ensemble sentiment analysis engine. Using Afinn, GloVe, and Word2Vec, the sentiment analysis engine labels statements as “positive,” “negative,” or “neutral.” These TSPs teach a 1D CNN to identify fake profiles with high accuracy. The true-human behavioral theory influenced the model, which predicts that false users’ periodic data will always converge. User perspectives are accounted for in the model. Therefore, the model achieves a 5.9%, 4.5%, and 3.2% improvement over state-of-the-art techniques for identifying bogus profiles, respectively. Multimodal social media interfaces benefit from this method.
