AI-102 Practice Question: Implement natural language processing solutions
You have a custom Named Entity Recognition (NER) model trained using Azure AI Language. The model is performing poorly on new data. You need to improve its accuracy. Which action should you take first?
⚠ Common exam trap
The trap here is that candidates often jump to hyperparameter tuning (epochs) or model simplification (reducing entity types) as a quick fix, when the core issue is almost always insufficient or low-quality labeled data for specific entities, which is the first diagnostic step in any custom NER workflow.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Review the test set results and add more labeled examples for entities with low precision/recall.
The first step to improve a custom NER model's accuracy is to analyze the test set results to identify which entity types have low precision or recall, then add more labeled examples for those specific entities. This targeted data augmentation addresses the root cause of poor performance—insufficient or imbalanced training data—rather than blindly adjusting hyperparameters or reducing complexity.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Increase the training epochs.
Why it's wrong here
More epochs will not fix poor accuracy caused by insufficient or unrepresentative labelled training data, and may overfit. It is tempting because epoch count is a familiar tuning knob, and increasing it would be correct when a model is undertrained and validation loss is still falling.
- ✗
Retrain the model using the same training data.
Why it's wrong here
Retraining on identical data reproduces the same model weights, since Azure AI Language learns only from labelled examples; the poor performance stems from data that does not match production. It is tempting because retraining is the standard remedy when new labelled utterances are added, which would be correct here.
- ✓
Review the test set results and add more labeled examples for entities with low precision/recall.
Why this is correct
Inspecting test set metrics exposes which entities have low precision or recall, so additional labelled examples target the actual weaknesses. Retraining blindly or adding unrelated data would not address the specific accuracy deficit the model exhibits on new data.
- ✗
Reduce the number of entity types in the model.
Why it's wrong here
Removing entity types discards labels the model must still extract, reducing recall rather than fixing the underlying data gap. It is tempting when overlapping or ambiguous entity schemas cause confusion, and consolidating them would be correct if the stem showed mislabelled, conflicting categories.
Go deeper
Related to this question
About these practice questions
This AI-102 question is part of Courseiva's 761-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI-102 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI-102 exam.