计算机应用与软件2026，Vol.43Issue(5)：105-110,6.DOI:10.3969/j.issn.1000-386x.2026.05.014
用于社交媒体抑郁评估的对比学习方法研究
A COMPARATIVE LEARNING METHOD STUDY FOR DEPRESSION ASSESSMENT ON SOCIAL MEDIA
摘要
Abstract
Social media provides a new avenue for early detection of depression.However,there are still two challenges to overcome,namely,jointly considering textual and visual information and extracting common features related to depression from multimodal data.Therefore,a contrastive learning method is proposed for multi-modal depression assessment.This method encoded textual and visual data,extracted data features of different modalities using feature extraction branches,aligned and fused multimodal features through a multimodal fusion module,and utilized the approach of multi-task learning to assist the model in learning the common features related to depression in both textual and visual information.Experiments show that the proposed method helps improve the accuracy of depression assess-ment.关键词
社交媒体/抑郁评估/对比学习/神经网络/多模态Key words
Social media/Depression assessment/Comparative learning/Neural networks/Multi-modal分类
信息技术与安全科学引用本文复制引用
陈晔,杨长春,王彭,杨森..用于社交媒体抑郁评估的对比学习方法研究[J].计算机应用与软件,2026,43(5):105-110,6.基金项目
江苏省未来网络科研基金项目(FNSRFP-2021-YB-36) （FNSRFP-2021-YB-36）
2022年度研究生培养创新工程项目(YPC22020259) （YPC22020259）
江苏省现代教育技术研究2021年度立项课题(2021-R-88294). （2021-R-88294）
