JobAds Ground Truth Dataset (FWF Project P35783)

The file contains manuallty annotated coordinates of 14 985 job advertisements in 29 different newspaper titles from ANNO corpus. Each line contains a link to a page in the ANNO corpus and coordinates of all job advertisements on that page.

Each job ad is labeled as one of the following categories:
- job_offer for job offers,
- job_search for job searches,
- service_offer for somebody offering services,
- vermittlung for job ads from mediation offices,
- heading for a heading indicating that a job ads sections starts,
- indicators for more general headings indicating that an ads section starts.

Ground truth was created as part of the JobAds (FWF P35783) and published as part of the conference paper "Who Advertises in Newspapers? Data Criticism in Mining Historical Job Ads" presented at the CHR2024 conference.

We thank the Austrian National Library for providing data from the ANNO corpus.