Decoding the Cauzin Softstrip: a case study in extracting information from old media

Area of Science:

• Computer Science
• Digital Curation
• Artificial Intelligence

Background:

• Archived digital content on obsolete media is inaccessible if unreadable.
• Cauzin Softstrips, a 1985 two-dimensional barcode, encoded digital data on physical media like paper.
• Physical readers for Softstrips are rare and becoming non-functional, posing a risk to data preservation.

Purpose of the Study:

• To develop a digital method for extracting data from Cauzin Softstrips.
• To make historical digital information encoded on Softstrips accessible without physical readers.
• To demonstrate the application of deep learning in digital archaeology and data recovery.

Main Methods:

• A convolutional neural network was employed to create a digital Softstrip reader.
• A corpus of 1229 Softstrips was used to train and test the decoding strategy.
• An additional set of 117 Softstrips was acquired for further validation.

Main Results:

• The developed decoding strategy achieved over 91% accuracy on the initial corpus.
• Accuracy increased to 99% for Softstrips produced under controlled conditions.
• Nearly 95% of an independently acquired set of Softstrips were decoded successfully.

Conclusions:

• Deep learning offers an accessible solution for digital data recovery from obsolete formats.
• The developed digital reader effectively bypasses the need for scarce physical Cauzin Softstrip readers.
• This research highlights the potential of AI in digital curation and the preservation of historical digital information.