Describe features of Natural Language Processing workloads on Azure practice questions

Language Understanding (LUIS)

Why wrong: LUIS is designed for understanding intents and extracting entities from short conversational phrases, not for batch processing of long, unstructured medical notes.

Translator Text

Why wrong: Translator Text performs language translation, not extraction of structured data elements from text.

Speech

Why wrong: The Speech service converts audio to text and performs other speech-related tasks; it does not extract entities from written text.

Speech to Text and Text Analytics API (standard)

Why wrong: The standard Text Analytics API can extract general entities like names and locations but does not have built-in medical entity recognition.

Translator Text and Language Understanding (LUIS)

Why wrong: Translator Text translates between languages, and LUIS extracts intent and entities from utterances. Neither service transcribes audio nor specializes in medical extraction.

Speaker Recognition and Question Answering

Why wrong: Speaker Recognition identifies who is speaking, and Question Answering provides answers from a knowledge base. They do not perform transcription or medical entity extraction.

Text Analytics

Why wrong: Text Analytics provides sentiment analysis, key phrase extraction, and entity recognition, but not intent recognition.

Translator Text

Why wrong: Translator Text is for language translation, not intent recognition.

Speech-to-text

Why wrong: Speech-to-text converts spoken language to text but does not understand intent.

Sentiment Analysis

Why wrong: Sentiment analysis evaluates the emotional tone of the text (positive, negative, neutral). It does not list the topics discussed.

Named Entity Recognition

Why wrong: Named Entity Recognition identifies specific entities such as names of people, organizations, locations, dates, etc. It does not summarize the general topics of the article.

Language Detection

Why wrong: Language detection identifies the language in which the text is written. It does not provide any information about the content or topics.

Sentiment Analysis

Why wrong: Sentiment Analysis determines the positive/negative/neutral sentiment of text but does not create a knowledge base for question answering.

Key Phrase Extraction

Why wrong: Key Phrase Extraction identifies important phrases in text but cannot answer specific questions based on document content.

Language Detection

Why wrong: Language Detection identifies the language in which text is written, but does not provide answers to questions.

Sentiment analysis

Why wrong: Sentiment analysis determines the emotional tone (positive, negative, neutral) of text, but does not extract specific topics or aspects.

Named entity recognition

Why wrong: Named entity recognition identifies predefined categories such as person, organization, location, but does not typically extract general topic phrases like 'battery life'.

Language detection

Why wrong: Language detection identifies the language in which the text is written, not the topics or themes discussed.

Named Entity Recognition (NER)

Why wrong: NER extracts entities like names of people or organizations, but it does not assign a single category to the whole document.

Key phrase extraction

Why wrong: Key phrase extraction identifies important phrases but does not categorize documents into predefined categories.

Conversational Language Understanding (CLU)

Why wrong: CLU is designed for understanding intents and entities from conversational messages, not for classifying full documents into categories.

Sentiment Analysis

Why wrong: Sentiment analysis detects positive, negative, or neutral sentiment, not custom entities.

Key Phrase Extraction

Why wrong: Key phrase extraction identifies important phrases but does not allow custom entity definitions.

Conversational Language Understanding

Why wrong: CLU is designed for conversational contexts (intents and entities) and is not optimized for document-level custom entity extraction.

Key Phrase Extraction

Why wrong: Key phrase extraction identifies important phrases but does not associate sentiment with specific aspects. It would return 'battery life' and 'design' but not the sentiment for each.

Sentiment Analysis

Why wrong: Sentiment analysis provides an overall positive/negative score for a document or sentence but does not identify which aspects correspond to which sentiment.

Named Entity Recognition (NER)

Why wrong: NER extracts entities like people, organizations, and locations. Product aspects like 'battery life' are not standard entities in the built-in model.

Sentiment Analysis

Why wrong: Sentiment Analysis measures the emotional tone of text, not the extraction of key phrases or topics.

Named Entity Recognition (NER)

Why wrong: NER identifies specific entities like people, organizations, dates, and locations, but not general descriptive phrases about issues.

Language Detection

Why wrong: Language Detection identifies the language of the text, which is not useful for extracting topics or issues.

Azure AI Translator and Azure Anomaly Detector

Why wrong: Anomaly Detector is for identifying unusual patterns in time-series data, not for language translation or sentiment analysis.

Azure AI Speech and Azure AI Language (Text Analytics)

Why wrong: Speech services are for audio-to-text or text-to-speech, but the input is already text; translation would still be missing.

Azure AI Vision and Azure AI Language (Text Analytics)

Why wrong: Vision services analyze images, not text; translation is not provided by Vision.

Key Phrase Extraction

Why wrong: Key Phrase Extraction identifies important terms in text but does not have a specialized medical model; it yields general phrases, not structured medical entities.

Named Entity Recognition (NER)

Why wrong: Standard NER recognizes general entities like people, organizations, and locations, but it is not trained to extract medical-specific entities like diseases or medications.

Custom Text Classification

Why wrong: Custom Text Classification would require the user to provide labeled training data and train a model, which is not a prebuilt capability aimed at medical entities.

Named entity recognition

Why wrong: Named entity recognition extracts specific entities (e.g., person names, locations) but not general important phrases.

Sentiment analysis

Why wrong: Sentiment analysis evaluates the emotional tone (positive, negative, neutral), not the key topics.

Language detection

Why wrong: Language detection identifies the language of the text, not the meaningful phrases within it.

Key Phrase Extraction

Why wrong: Key Phrase Extraction identifies the main talking points in text but does not evaluate whether the sentiment is positive or negative.

Named Entity Recognition

Why wrong: Named Entity Recognition extracts entities like names, dates, and organizations, but it does not provide sentiment information.

Language Detection

Why wrong: Language Detection identifies the language of the input text, not the sentiment expressed in it.

Custom Named Entity Recognition

Why wrong: Custom NER requires users to train a model with labeled data, which the scenario explicitly says they do not want to do.

Key Phrase Extraction

Why wrong: Key Phrase Extraction identifies general important phrases, not domain-specific medical entities.

Sentiment Analysis

Why wrong: Sentiment analysis determines the emotional tone of text, not extracting medical entities.

Sentiment analysis

Why wrong: Sentiment analysis determines the overall emotional tone of text, not specific entity extraction.

Key phrase extraction

Why wrong: Key phrase extraction returns important phrases but does not classify them into categories like person or organization.

Language detection

Why wrong: Language detection identifies the language of the text, not specific entities.

Sentiment Analysis

Why wrong: Sentiment Analysis detects the sentiment (positive, negative, neutral) of text but does not extract answers from a knowledge base.

Key Phrase Extraction

Why wrong: Key Phrase Extraction identifies important words or phrases in text, but it does not answer questions or retrieve pre-defined answers.

Text Analytics for Health

Why wrong: Text Analytics for Health is specialized for extracting medical entities and relationships from clinical text, not for general question answering.

Sentiment Analysis

Why wrong: Sentiment Analysis determines whether text is positive, negative, or neutral, not for extracting sensitive entities.

Key Phrase Extraction

Why wrong: Key Phrase Extraction pulls out important terms but does not specifically target or categorize sensitive personal information.

Language Detection

Why wrong: Language Detection identifies the language of the document but does not extract any entities.

Key Phrase Extraction

Why wrong: Key Phrase Extraction identifies important words and phrases in general text, but it is not specialized for medical terminology and does not provide the structured medical entity recognition needed.

Named Entity Recognition (NER)

Why wrong: Standard NER can identify general categories like person, location, and organization, but it is not pre-trained with medical-specific entities like diseases or medications.

Sentiment Analysis

Why wrong: Sentiment Analysis determines whether text is positive, negative, or neutral, not for extracting medical entities.

Key phrase extraction

Why wrong: Key phrase extraction returns individual important words or short phrases, not a coherent summary of key points.

Named entity recognition

Why wrong: Named entity recognition identifies entities like people, places, and organizations but does not produce a summary.

Sentiment analysis

Why wrong: Sentiment analysis detects the emotional tone (positive, negative, neutral), not key content or summary.