Microsoft AI-102 - Designing and Implementing a Microsoft Azure AI Solution

To maximize answer accuracy (reduce randomness and hallucinations) with Azure OpenAI GPT-3.5, you should lower the sampling randomness. The primary control for this is temperature:

Temperature controls how “creative” or random the model is.

Lower values → more deterministic, focused, and conservative outputs.

temperature = 0 makes the model pick the highest-probability tokens, which typically yields the most accurate and repeatable responses for transactional bots like booking/reservations.

Top P (nucleus sampling) is an alternative to temperature. Microsoft recommends using one or the other. While reducing Top P also narrows randomness, the clearest and most common approach for maximizing accuracy is setting temperature to 0. Setting Top P to 0 is generally not recommended and may be invalid depending on the SDK.

System message tuning (Option D) helps with tone, role, and policy adherence, but does not itself guarantee factual accuracy if decoding randomness remains high.

Including the travel agent’s database (Option A) would require Azure OpenAI on your data or a RAG pattern. That can improve grounding of facts, but it’s not a model parameter change and is outside the scope of simply configuring GPT-3.5; moreover, accuracy still benefits from low temperature even when using your own data.

Therefore, the best single action here is: set Temperature to 0.

Microsoft Azure AI Solution References

Azure OpenAI Service – Chat Completions parameters (Temperature and Top P): “Lower values like 0.2 will make the output more focused and deterministic… use either temperature or top_p but generally not both.”https://learn.microsoft.com/azure/ai-services/openai/reference#chat-completions

Prompt engineering with Azure OpenAI – controlling randomness and determinism:https://learn.microsoft.com/azure/ai-services/openai/concepts/prompt-engineering

Azure OpenAI on your data (grounding with enterprise data/RAG):https://learn.microsoft.com/azure/ai-services/openai/concepts/use-your-data

=========================

The requirement is:

"Measure the public perception of your brand on social media using natural language processing (NLP)."

A. Content Moderator

Designed to filter offensive or inappropriate content in text, images, and videos.

Not suitable for sentiment analysis or measuring perception.

B. Form Recognizer

Extracts structured data from documents (invoices, receipts, forms).

Not used for analyzing social media posts.

C. Computer Vision

Analyzes images and videos for object detection, image classification, OCR, etc.

Not used for text-based sentiment analysis.

D. Language service ✅

Provides natural language processing (NLP) capabilities such as sentiment analysis, opinion mining, key phrase extraction, entity recognition, and language detection.

Perfect fit for analyzing public perception from social media posts.

The output will include the following words: our and included. → No

The output will include the following words: Paris, Eiffel, and Tower. → Yes

The function will output all the key phrases from the input string to the console. → No

Comprehensive Detailed ExplanationLet’s examine the function:

static void GetKeyWords(TextAnalyticsClient textAnalyticsClient, string text)

{

var response = textAnalyticsClient.RecognizeEntities(text);

Console.WriteLine("Key words:");

foreach (CategorizedEntity entity in response.Value)

{

Console.WriteLine($"{entity.Text}");

}

}

The function uses RecognizeEntities from the Azure Text Analytics SDK.

RecognizeEntities detects and categorizes named entities in text (such as locations, people, organizations, dates, etc.).

It does not extract all words or key phrases.

For extracting key phrases, you would use ExtractKeyPhrases, not RecognizeEntities.

"The output will include the following words: our and included."

These are common words, not entities. They will not appear.

Answer: No

"The output will include the following words: Paris, Eiffel, and Tower."

These are named entities (location, landmark). They will be recognized and returned.

Answer: Yes

"The function will output all the key phrases from the input string to the console."

No. The function extracts entities, not key phrases. Key phrases would require ExtractKeyPhrases.

Answer: No

Evaluating the statements:

Database transactions follow the ACID properties:

Atomicity → Ensures that each transaction is treated as a single “all-or-nothing” unit of work. Either the entire transaction is completed, or none of it is applied. This guarantees that transactions are executed only once and rolled back completely if they fail.

Consistency → Ensures that a transaction brings the database from one valid state to another, preserving integrity constraints.

Isolation → Ensures that concurrent transactions do not interfere with each other.

Durability → Ensures that once a transaction is committed, it remains permanent even in case of system failure.

Thus, the correct property here is Atomicity.

Microsoft Reference: ACID properties in databases

Speech-to-text converts the caller’s spoken French/German audio into text.

The text must then be translated to English.

Finally, Text-to-speech converts the translated text into spoken English for playback.

Therefore, the combination of Translator + Text-to-speech is required for the replay step.

Comprehensive Explanation

Speech-to-text in Azure AI Speech

Azure AI Translator

Text-to-speech in Azure AI Speech

Microsoft References

Final Answer:

To capture messages → Speech-to-text

To replay messages → Text-to-speech and Translator

Comprehensive Detailed ExplanationThere are two main types of data workloads: Batch and Streaming.

Batch Workload

Processes data in large groups (batches) at scheduled intervals.

Common when real-time is not required.

Examples: daily sales reports, hourly data refreshes, periodic ETL (Extract, Transform, Load).

Streaming Workload

Processes data in real time or near real time as it arrives.

Used for scenarios where immediate insights or actions are required.

Examples: fraud detection, IoT sensor monitoring, live purchase tracking.

“Data for a product catalog will be loaded every 12 hours to a data warehouse.”

This is scheduled, not continuous. → Batch

“Thousands of data sets per second for online purchases will be loaded into a data warehouse in real time.”

This requires real-time processing. → Streaming

“Updates to inventory data will be loaded to a data warehouse every 1 million transactions.”

Data is grouped and loaded after reaching a threshold (1 million). → Batch

Applying to the scenarios:

Correct Matches:

Product catalog (12 hours) → Batch

Online purchases (real time) → Streaming

Inventory data (1 million transactions) → Batch

Batch processing vs. Stream processing in Azure

Azure Stream Analytics overview

Data processing approaches: Batch vs Real-time

Microsoft References

For a social media messaging app that must identify the language of text messages in real time, the most direct SDK is the Azure AI Translator – Text Translation SDK. Its client libraries (e.g., Azure.AI.Translation.Text in .NET) include a language detection operation that returns the detected language code and a confidence score, which is exactly what you need for real-time text scenarios.

Other options are not a fit:

Microsoft.CognitiveServices.Speech is the Speech SDK for speech-to-text, TTS, and speech translation (audio), not text-only language detection.

Azure.AI.Translation.Document targets asynchronous document translation workflows, not short text message detection. Microsoft Learn

Azure.AI.Translation.Speech isn’t the package used; speech translation is covered by the Speech SDK package above. Microsoft Learn

Microsoft References

Text Translation SDK overview (supports language detection): Azure AI Translator. Microsoft Learn

DetectedLanguage in Azure.AI.Translation.Text (returned by detect operation). Microsoft Learn

Quickstart: Translator text client libraries. Microsoft Learn

Install/use the Speech SDK (contrast with text scenario). Microsoft Learn

Document Translation SDK overview (not applicable to short text detection). Microsoft Learn

Scenario: All planned projects must support English, French, and Portuguese.

If a bot uses multiple LUIS models and QnA Maker knowledge bases (knowledge bases), you can use the Dispatch tool to determine which LUIS model or QnA Maker knowledge base best matches the user input. The dispatch tool does this by creating a single LUIS app to route user input to the correct model.