Knowledge Extraction from Unstructured Texts

OpenIE (Open Information Extraction) is a tool that filters and normalizes raw text between entities to obtain open-domain relations. Distant supervision, also called weak supervision, is when we wish to extract relations from a text and that an initial KG is available, as a seed, for the triplet extraction task. If we have a lot of texts and a big enough KG (with the same entities as the ones in the texts) we can learn a mapping from raw text to fixed-schema relations of the KG.

Source: blog.heuritech.com