Challenges:
Recognize and rename fields to concept references automatically, via ML.
- Create a large number of these normalization templates for a lot of data sources.
- Machine-learn the field-names, and corresponding concepts, so that the data gets normalized automatically, and ready for analysts.
- Result:
- we shall have a universal data normalizer, which given any data, is able to automatically:
- assign proper fields and value types to it.
- identify their position in the nested hierarchies
- understand their ontological meaning in contexts of their positions