Entities are additional elements of structured data which can be extracted from within the verbatims. Entities include data such as monetary quantities, dates, currency codes, organisations, people, email addresses, URLs, as well as many other industry specific categories.
The below screenshot shows a verbatim containing three predicted entities; two monetary quantities and a policy number:
An example email verbatim sent to an insurance underwriting mailbox containing structured data entities: two monetary quantities and a policy number
Much like labels, predicted entities can be accepted, rejected, or assigned by highlighting a string of text and choosing the correct entity from the list in the modal (see here for how). Both of these actions will provide training signals to the entity extraction model, which will improve its understanding of that entity type.
Enabling entity extraction and selecting the entities to extract are confirmed either during the creation of the dataset or via the settings section in the Dataset settings page.
Previous: Datasets | Next: Labels (predictions, confidence levels, hierarchy, etc.)