Learned index structures are a term from the fields of Artificial Intelligence, Big Data and Smart Data, and Digital Transformation. They describe a modern method for searching large volumes of data faster and more efficiently by using Artificial Intelligence to optimise so-called „data indexes“.
Traditionally, index structures – which are special database rules for fast searching – were programmed by experts. With learned index structures, an artificial intelligence takes over, learning from existing data and then independently finding better ways to store and retrieve data.
A practical example: In a company, millions of customer data records are stored. For example, searching for all buyers who spent over 500 Euros in the last month can take a long time. However, if the company uses learned index structures, artificial intelligence can recognise typical search patterns and organise the data so that the search works much faster.
Learned index structures therefore help to make large databases more efficient – especially where a lot of information is needed at lightning speed. This is an important building block for modern, data-driven companies.
Deutsch 
English (UK)