The term "training data curation" comes from the fields of artificial intelligence, big data, smart data and digital transformation. If artificial intelligence is to learn or perform tasks independently, it needs examples - known as training data. Training data curation means that this sample data is carefully selected, checked and sorted. This is important to ensure that the AI works reliably and accurately.
Imagine a company wants to train an AI to recognise tumours on X-ray images. Thousands of images are collected for this purpose. During training data curation, experts check which images are usable, sort out faulty or irrelevant images and ensure that the data is diverse enough. This prevents the AI from drawing false conclusions or only being trained on a few patterns.
Without good training data curation, automated systems could make incorrect decisions or work inaccurately. This preliminary work is therefore crucial, especially in sensitive areas such as medicine, finance or autonomous vehicles. Companies investing in artificial intelligence should not underestimate the importance and effort of training data curation in order to obtain reliable results.
Deutsch 
English (UK)