Big Data Analysis Based On Lex Sense Feature Selection Using Deep Featured Neural Classification For Identifying Fake News In Social Tags

Abstract

newline xv newlineABSTRACT newlineWith the widespread use of social network services, fake news which is a way newlineof disguising false information as legitimate information has become a big social newlineissue. Fake news contains misleading information that could be checked. To resolve newlinethe problem in first approach a Lex-Sense Feature Selection (LSFS) and Grid-Based newlineTransfer Learning (GBTL) are implemented to predict the truth information. The newlineSpectral Recurrent Neural Classification (SRNN) is optimized to identify the social newlinemedia truth information category to reduce the rumors to create awareness newlinerecommendations. Initially, the preprocessing is carried out to apply the Natural newlinelanguage processing filters to make redundant data. Further, the Lexi terms are newlineanalyzed with a syntactic sentiment analyzer to predict maximum non-related tags newlinefrom affected real terms from positive comments. Further, Grid-based Transfer newlineLearning (GBTL) compares the adverbs, preposition terms to compare the senti-word newlinenet to observe the subjectivity weight of the sentence. Then, the selected feature newlineweights are trained with soft-max activation function to create the polarity weight and newlineclassified with the recurrent neural network. newlineThe second approach proposes a Hyper-Lexi Phantom Feature Selection newline(HLPFS) based on Topic Vector Lex Sense (TVLS) be optimized with a convolution newlineneural network to predict the fake tweets information in social media. Initially, the newlinepreprocessing was carried to fake redundant data to predict the topic-relevant model newlineusing the Topic Vector Summarization Rate (TVSR). This selects the geometric newlinefeatures of discussion forums content. Based on that topic relevance process, the newlineReliable Entity Content Hit Rate (RECHR) observe positive recurrent information newlinedifference between the tweets lexical weights. Hyper-Lexi Phantom Feature Selection newline(HLPFS) selects the importance of fake difference metrics among reliable terms to newlinereduce dimension based on the entity rate. Then selected features are classified with newlineadaptive decision in convolution