Open Big Data

1. Filename: Unique identifier of this advertisement
2. URLs TheEconomistPageScans: comma separated list of URLs to JPG image files of scanned The Economist pages containing this ad. For multi page ads this can be multiple URLs.
3. Date of Issue: Date of The Economist issue (Years-Month-Day)
4. Bounding Box relative X1: Left-top coordinate of a rectangle identifying the ad on the page, relative to the pixel coordinates of the image from column 2 (“URLs …”). Multiply this value by the width of the image to get the absolute x coordinate. If the ad is a multi page ad, the images from column 2 have to be horizontally concatenated first.
5. Bounding Box relative Y1: Left-top coordinate
6. Bounding Box relative X2: Right-bottom coordinate
7. Bounding Box relative Y2: Right-bottom coordinate
8. Brand: Brand name of advertiser
9. Brand is generic (e.g. ‘Notices’): If “True” then this ad doesn’t represent a single brand, but a category of ad-like content. Most common categories are “Notices”, “Appointments”, “Courses”.
10. OCR GoogleVision: Advertisement text, based on text recognition using Google Vision API (2021) of the full ad image.
11. Text Class GoogleVision: Based on the OCR text the ad was classified using GoogleVision API (2021). See full list of categories. This column contains a JSON string with a list of text classes and their class probabilities.

Authors

Kluge, S., Gehrmann, L., Stahl, F., Knäble, M., Nadj, M., Maedche, A.

Languages

English

Funding / Grants

DFG

License

Creative Commons Attribution 4.0 International

DOI

https://doi.org/10.7801/382