Improving question-answering models that use data from tables
5-14% improvement. Amazon just cracked the code on AI reading tables.

Why it matters
Table-based question answering is critical for enterprise AI applications that need to extract insights from structured business data. This technical breakthrough could significantly improve how AI systems interact with spreadsheets, databases, and financial reports.
The key facts
3 to know5% to 14% improvement across five evaluation metrics
Novel pretraining method for table-based question answering
Published by Amazon Science research team
Go to the source
Amazon Scienceamazon.science
Publisher excerpt: Novel pretraining method enables increases of 5% to 14% on five different evaluation metrics.

