JobAds: From Digitized Newspapers to Economic Analysis

The JobAds Project (FWF P35783) explores the Austrian labor market from 1850 to 1950 by analyzing job advertisements in digitized newspapers. The research is divided into two phases: converting data into machine-readable text and conducting economic analysis. The first phase involved identifying pages with job ads in the ANNO corpus, their segmentation, converting them into machine-readable text, and refining this text through post-correction steps, as well as the manual annotation of segments and creating gold standard by correcting the automatically recognized text. Now, the project focuses on extracting job advertisements, analyzing their structure and labor market trends. This requires understanding the historical and economic context of the job ads, including metafacts such as the role of private agencies or the political orientation of newspapers. Future tasks involve using BERT topic modeling to identify key topics in the corpus and to gain insights into historical job market dynamics.