基于上下文语义的社交网络用户人格预测  

Social network user′s personality prediction based on context semantics

作　　者：王江晴[1] 陈思敏 刘晶[1] 孙翀[1] 毕建权 WANG Jiangqing;CHEN Simin;LIU Jing;SUN Chong;BI Jianquan(College of Computer Science,South-Central University for Nationalities,Hubei Provincial Engineering Research Center for Intelligent Management of Manufacturing Enterprises,Wuhan 430074,China)

机构地区：[1]中南民族大学计算机科学学院,湖北省制造企业智能管理工程技术研究中心,武汉430074

出　　处：《中南民族大学学报（自然科学版）》2020年第3期289-294,共6页Journal of South-Central University for Nationalities：Natural Science Edition

基　　金：国家科技支撑计划项目(2015BAD29B01);湖北省技术创新专项重大项目(2019ABA101);湖北省自然科学基金资助项目(2017CFC886);中央高校基本科研业务费专项资金资助项目(YZY18002);中南民族大学研究生学术创新基金资助项目(2019sycxjj122)。

摘　　要：在利用文本信息预测用户大五人格的普遍方法中,对于文本特征的提取未充分考虑上下文语义信息,存在对语义特征提取不够精准的问题.针对该问题,提出了一种结合深度学习与上下文语义的方法:在TF-IDF中加入单词的上下文语义信息来计算单词权值,然后结合基于文本的卷积神经网络模型和由单词权值构成的上下文语义特征向量进行用户大五人格预测.实验数据使用Facebook中myPersonality应用的用户社交记录,实验结果表明:将文本上下文语义加入到深度学习预测模型后,人格预测的准确率有所提高.In the general methods of using text information to predict user′s Big Five personality,the extraction of text features does not fully consider context semantics information,and there is a problem that semantic features are not extracted accurately enough.Aiming at this problem,a method combining deep learning and context semantics is proposed.The context semantics information of words is added to TF-IDF to calculate word weight,and then combining text-based convolutional neural network model and context semantics feature vector composed of word weight to predict user′s Big Five personality.The experimental data uses the user′s social record of the myPersonality application in Facebook,and the experimental result shows that accuracy of user′s Big Five personality prediction is improved after the text context semantics is added to the deep learning prediction model.

关 键 词：人格预测 大五人格 上下文语义 深度学习 社交网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]