基于情绪计算的GIS硕士面试成绩预测  

Interview Scores Prediction for GIS Master Based on Emotional Computing

作　　者：费腾[1] 付康钰 卞萌 杨妙玲[1] FEI Teng;FU Kangyu;BIAN Meng;YANG Miaoling(School of Resource and Environmental Science,Wuhan University,Wuhan 430079,China;School of Remote Sensing and Information Engineering,Wuhan University,Wuhan 430079,China)

机构地区：[1]武汉大学资源与环境科学学院,湖北武汉430079 [2]武汉大学遥感信息工程学院,湖北武汉430079

出　　处：《地理空间信息》2024年第4期128-132,共5页Geospatial Information

基　　金：国家自然科学基金资助项目(42271476)。

摘　　要：针对组织GIS专业学生面试带来人力资源与时间成本大的问题,本研究基于真实面试的视频数据和专家综合打分数据,通过人脸情绪识别与构建特征工程,使用包括偏最小二乘回归、支持向量回归、随机森林回归和梯度提升回归树的4种机器学习模型进行回归模型构建,探究情绪特征与面试成绩的关系。结果表明偏最小二乘回归模型的稳定性和泛化能力在本应用中更好,模型的决定系数为0.486。进一步分析发现,基本情绪中,我们定义的“负面情绪”与面试成绩呈显著负相关;而中性情绪、积极情绪与面试成绩呈显著正相关。本研究为GIS人才面试评估的自动化做出了有益的探索。In response to the problem of organizing interviews for GIS specialties that bring large human resources and time costs,based on real interview video data and expert scoring data,we explored the association between emotional features and interview scores through face emotion recognition and construction feature engineering using four machine learning models including partial least squares regression,support vector regression,random forest regression and gradient boosting regression trees for regression model construction.The results show that the stability and generalization ability of partial least squares regression model are better in this application,and the determination coefficient of model is 0.486.Further analysis reveals that among the basic emotions,our definition of“negative emotions”is significantly negatively correlated with interview performance,while neutral emotions and positive emotions are significantly positively correlated with interview performance.This study can make a meaningful exploration for the automation of GIS talent interview assessment.

关 键 词：特征工程 情绪提取 机器学习 GIS专业 

分 类 号：P208[天文地球—地图制图学与地理信息工程]