摘要
随着新冠肺炎疫情肆虐全球,抑郁症患病率大幅增加,高患病率和低就诊率成为抑郁症面临的两大挑战。人工智能与机器学习的快速发展为精准识别出抑郁倾向人群奠定了基础。文中借助微博平台,建立用户画像对用户进行抑郁倾向识别,构建基于用户画像的TCNN⁃GRU⁃PR融合识别模型。首先,利用TF⁃IDE算法扩充基础种子词,构建抑郁倾向情绪词典;然后,采用TCNN⁃GRU模型进行情绪特征提取识别,引入PageRank算法从社交网络维度进行再次识别;最后,将机器学习得出的情绪标签概率值与PR值加权求和,判断抑郁倾向程度。实验结果表明,TCNN⁃GRU⁃PR模型结合文本识别和社交网络识别两个维度能准确识别出情绪状态和抑郁程度,对于抑郁倾向人群的早期识别与干预治疗有重要意义。
With the COVID⁃19 sweeping the world,the prevalence of depression has increased significantly.High prevalence and low consultation rate have become two major challenges for depression.The rapid development of artificial intelligence and machine learning has laid the foundation for accurately identifying people prone to depression.With the help of Weibo platform,a user portrait is established to identify depression tendencies of users,and a TCNN⁃GRU⁃PR fusion recognition model based on user portrait is constructed.TF⁃IDE algorithm is used to expand the basic seed words and construct a depression tendency emotion dictionary.TCNN⁃GRU model is used for emotional feature extraction and recognition,and PageRank algorithm is introduced for recognition from the social network dimension.The weighted summation of the probability value of the emotion tag obtained from machine learning and the PR value is conducted to determine the degree of depression tendency.The experimental results show that TCNN⁃GRU⁃PR model combined with text recognition and social network recognition can accurately identify emotional status and depression degree,which is important for early identification and intervention treatment of depression prone populations.
作者
黄承宁
徐新
朱玉全
HUANG Chengning;XU Xin;ZHU Yuquan(Nanjing Technology University Pujiang Institute,Nanjing 211222,China;Jiangsu University,Zhenjiang 212013,China)
出处
《现代电子技术》
2023年第10期143-148,共6页
Modern Electronics Technique
基金
国家自然科学基金项目(61702229)
江苏省高等学校自然科学研究项目(18KJD520001)
南京工业大学浦江学院优秀青年骨干教师工程(南工大浦校[2021]73号)
南京工业大学浦江学院人才培养工程计划项目(njpji2019⁃1⁃04)。
