Table of Contents
Which NLP entity type uses labeled data to identify complex entities?
Get ready for your AI-900 exam with a detailed look at Azure AI’s natural language processing (NLP) entities. Understand why a machine-learned entity is the correct type for identifying complex, context-dependent information like names and addresses by using labeled training data.
Question
Which type of natural language processing (NLP) entity in Azure AI is used to identify complex entities, such as names or addresses, by leveraging algorithms trained on labeled data?
A. Regular expression
B. Machine-learned
C. Prebuilt entity
D. Pattern.any
Answer
B. Machine-learned
Explanation
The correct type of NLP entity for identifying complex items like names or addresses is B. Machine-learned. This entity type is specifically designed to recognize entities based on the context provided by labeled examples in your training data.
Understanding Machine-Learned Entities
A machine-learned entity is a component in Azure AI Language (and its predecessor, LUIS) that learns to identify information from the way it’s used in sentences. Unlike entities that rely on fixed lists or rigid patterns, a machine-learned entity uses algorithms to learn the contextual clues surrounding the entity. For complex and highly variable data like personal names or street addresses, this is the most effective approach. The model learns from the labeled examples you provide what a name “looks like” in various sentence structures, enabling it to identify new, unseen names in different contexts.
The Training Process
- Labeling: You provide a set of example sentences and manually tag (label) the words that constitute the entity you want to extract (e.g., highlighting “John Smith” as a PersonName entity).
- Training: The AI model analyzes these examples, learning the features, word choices, and sentence positions that indicate the presence of the entity.
- Prediction: Once trained, the model can predict the presence of the entity in new, unlabeled text by applying the patterns it has learned.
Why Other Options Are Incorrect
- Regular expression: This method is too rigid for identifying names or addresses. The variations in names and address formats worldwide are too vast to be captured by a single, manageable pattern.
- Prebuilt entity: While Azure offers prebuilt entities for common types like personName and address, these prebuilt models are themselves sophisticated machine-learned models trained by Microsoft. The question asks for the type of entity used for such complex tasks, which is fundamentally machine-learned.
- Pattern.any: This entity is designed for data that follows a template or a consistent but flexible structure, like an invoice number or a product code. It is not suitable for entities like names that lack a predictable structural pattern.
Microsoft Azure AI Fundamentals AI-900 certification exam practice question and answer (Q&A) dump with detail explanation and reference available free, helpful to pass the Microsoft Azure AI Fundamentals AI-900 exam and earn Microsoft Azure AI Fundamentals AI-900 certification.