AI-900 Practice Question: Describe features of Natural Language Processing workloads on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Renaming standard NER entity types to match your organisation's terminology

  Why it's wrong here

  Renaming standard NER entity types is just a display or configuration change where you alias existing categories like 'Person' to a different label in your application. Custom NER does not rename pre-built types; it trains the model to recognise entirely new entity types defined by your labels. These new types may have no counterpart in the standard NER taxonomy, so the model must learn their semantics from annotated samples rather than from a simple name swap.

• ✓

  Training a model to recognise domain-specific entities not covered by pre-built NER

  Why this is correct

  Custom NER in Azure AI Language is a machine-learning model trained on your own labelled examples to extract domain-specific entities that the pre-built NER does not recognise, such as legal clause references, medical device names, or proprietary product codes. The pre-built NER only handles generic categories like Person, Location, and Organization, so any entity type unique to your industry requires a custom-trained model. This is the core definition of Custom NER: extending entity recognition beyond the out-of-the-box coverage.

• ✗

  A faster, lighter version of NER that uses simpler rules instead of machine learning

  Why it's wrong here

  Custom NER is not a rule-based, lighter alternative to standard NER; it is explicitly a machine-learning approach that learns patterns from labelled domain data. Traditional rule-based NER uses hand-crafted patterns or gazetteers, whereas Custom NER fine-tunes a pre-trained language model to generalise to your specific entity types. It is also not 'lighter' — it involves training, evaluation, and deployment of a custom model, which is more involved than applying simple rules.

• ✗

  Filtering NER outputs to return only the entity types relevant to your application

  Why it's wrong here

  Filtering NER outputs to only show certain entity types is simply a parameter or post-processing step applied after the pre-built NER runs, for example by selecting which categories appear in the API response. Custom NER, by contrast, trains an entirely new model to recognise entity types that the pre-built NER does not know at all. The training process creates a new entity schema, rather than merely limiting the set of pre-existing entity types returned.

Go deeper

Related to this question