ACM DocEng

The demand for accessible and reusable PDFs, optimized for screen readers and data extraction, is on the rise. Joseph explains a multi-year effort to automatically create accessible documents with LaTeX
#Accessibility #LaTeX #PDFInnovation #InclusiveTech
dl.acm.org/doi/10.1145/36…

The digitization of historical documents faces significant challenges due to OCR inaccuracies. James Zhang explores OpenAI's GPT models to improve post-OCR correction for texts from the Princeton Prosody Archive (1559-1928). #AI #OCR #Digitization
📕dl.acm.org/doi/10.1145/36…

Hrishikesh Kulkarni presents LexBoost, a document retrieval method that combines sparse and dense retrieval methods by leveraging a pre-build dense corpus graph. The result? Improved ranking performance with minimal computational overhead!

Full paper: dl.acm.org/doi/10.1145/36…

In an age where every problem is solved through machine-learning, Jean-Luc presents ZigZag, a machine-learning free algorithm that binarizes images which suffer from difficult lighting environments.
Read more: dl.acm.org/doi/10.1145/36…
Try the app: lnkd.in/gbXY_7jq

Sinan presents a new dataset bridging the gap between text descriptions and other document modalities used in the context of engineering design catalogs.
📝 Ready to dive deeper? Check out the paper here: dl.acm.org/doi/10.1145/36…

#document #engineering #design #dataset #NLP