SMPNet: An Algorithmic Framework for Loneliness Detection and Mitigation in Social Media

Venkatesh Sumukh, Yin Jack, Sng Grace, Aggarwal Raghav,Fan Weiguo, Huang Chengyue,Tong Ling

2023 IEEE MIT Undergraduate Research Technology Conference (URTC)（2023）

Cited 0|Views2

No score

Abstract

Loneliness is a growing problem in today's digital age. This study aimed to use NLP models for loneliness detection and prevention on social media platforms. Out of seventy-two combinations of eight models and nine preprocessing methods, SMPNet, made using LSA and MLP, with TFIDF performed best with 85% accuracy. Reddit and Discord bots were then created using SMPNet, able to detect loneliness, offer remedial resources, and alert moderators. The model was retrained on incorrect predictions, continuously improving its accuracy. The success of the model and bots means loneliness detection and prevention are very real and Implementable in social media environments.

Translated text

Key words

Social Media,Social Media Platforms,Incorrect Predictions,Natural Language Processing Models,Social Media Environment,Problem Today,Machine Learning,Learning Models,Learning Algorithms,Support Vector Machine,Twitter,Machine Learning Models,F1 Score,Deep Learning Models,K-nearest Neighbor,XGBoost,Text Classification,Lack Of Social Interaction,Emergence Of Social Media,Hate Speech,Tokenized,Positive Note,Instagram,Text Classification Tasks,Type Of Neural Network,Federated Learning,Neural Network

AI Read Science

Must-Reading Tree

Example

Generate MRT to find the research sequence of this paper

Chat Paper

Summary is being generated by the instructions you defined