A Person-Centric Prediction Model of Job Loss based on Social Media

摘要更好地了解除了公司的财务压力之外，哪些因素会预测失业，这符合商业和公共政策的利益。这项研究将应用大数据挖掘技术、社交网络和Twitter数据的统计分析，以确定预测失业的个人因素，而不仅仅是公司的商业和经济状况。这个话题对于管理和组织科学学科很重要，因为我们所知道的关于失业的大部分都是在社会和公司层面上的(例如工厂关闭)。该项目将使人们更好地了解工作性质的变化是如何影响人们的。研究人员正在采用一种基于大数据理论的混合方法，使用从流行的社交媒体推特获得的信息。本研究设计了两个研究：(1)利用语言和情绪分析从微观、中观和宏观角度识别Twitter数据中的失业预测因素；(2)分析Twitter用户的工作主题网络，推断网络指数作为失业预测因素。研究(1)和(2)中确定的预测指标集将单独和一起用作一系列推理模型的输入，这些模型将衡量每个信息领域所解释的工作损失的变异量。该项目将扩展基本的方法知识，展示如何将大数据挖掘与统计分析结合起来，从推文和社交媒体网络的内容中识别失业的多个预测因素。

AbstractIt is in the interests of business and public policy to better understand what factors predict job loss beyond a company's financial stress. This research will apply big data mining techniques, and social network and statistical analyses of Twitter data to identify personal factors that predict job loss, over and above a company's business and economic conditions. This topic is important for management and organization science disciplines because most of what we know about job loss is at the society and company level (such as plant closings). The project will lead to a better understanding of how the changing nature of work is affecting people.The researchers are taking a hybrid big data theory-based approach, using information derived from the popular social media Twitter. Two studies are devised: (1) Identify job loss predictors from micro, meso and macro perspectives in Twitter data by exploiting linguistic and sentiment analysis; (2) Analyze the job topic networks of Twitter users to infer network indices as predictors of job loss. The sets of predictors identified in studies (1) and (2) will be used independently and together as input to a cascade of inference models that will gauge the amount of variance in job loss explained by each of the information domains. This project will extend fundamental methodological knowledge by demonstrating how big data mining might be used in conjunction with statistical analysis to identify multiple predictors of job loss from the content of tweets and social media networks.