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

When implementing a backoff strategy, what return codes should you look for and then retry your request? -200 and 201. -404 and 501. -207 and 503.

207 and 503. 503 is the response when a request is throttled. 207 is the response when a single document in a batch fails.

You want to create a knowledge store that contains JSON representations of the indexed documents. What kind of projection should you define? -Object -File -Table

Object Object projections are JSON representations of an indexed document.

Which element of a dialog interprets user input in order to determine intent? -Language Generator -Trigger -Recognizer

Recognizer A recognizer interprets user input.

You are creating an index that includes a field named modified_date. You want to ensure that the modified_date field can be included in search results. Which attribute must you apply to the modified_date field in the index definition? -searchable -filterable -retrievable

retrievable To enable a field to be included in the results, you must make it retrievable.

Which element of a dialog formulates output that the bot sends to the user? -Language Generator -Trigger -Recognizer.

Language Generator A language generator formulates output from the bot to the user.

What is the effect of the smartCropping option when using Computer Vision to generate a thumbnail? -The aspect ratio of the original image is maintained. -The thumbnail is skewed to fit the specified proportions. -The region of interest is centered in the thumbnail.

The region of interest is centered in the thumbnail. Using smartCropping ensures that the region of interest is identified and centered in the thumbnail.

Which element of a dialog responds to a detected intent and initiates an appropriate response? -Language Generator -Trigger -Recognizer.

Trigger A trigger responds to a detected intent.

Your app must interpret a command such as "turn on the light" or "switch the light on". What do these phrases represent in a language model? -Intents. -Utterances. -Entities.

Utterances. Utterances are example phrases that indicate a specific intent.

You are viewing a metric for your Cognitive Services resource in a chart. You want to combine the chart with visualizations of other resources and data. What should you do? -Add the chart to a dashboard. -Share the chart. -Clone the chart.

Add the chart to a dashboard. -A dashboard enables you to combine visualizations from multiple resources.

How should you create an application that monitors the comments on your company's web site and flags any negative posts? -Use the Language service to extract key phrases. -Use the Language service to perform sentiment analysis of the comments. -Use the Language service to extract named entities from the comments.

Use the Language service to perform sentiment analysis of the comments. -sentiment analysis helps you determine if text is negative or positive.

How can you add a multi-turn context for a question in an existing knowledge base? -Add synonyms to the knowledge base. -Add alternative phrasing to the question. -Add a follow-up prompt to the question.

Add a follow-up prompt to the question. To add a multi-turn context to a question, define a follow-up prompt.

You've created a bot using a Bot Framework SDK template.Which tool should you use to test it locally? -Visual Studio Code -Bot Framework Emulator -Bot Framework Composer

Bot Framework Emulator Use the Bot Framework Emulator to test a bot locally.

Which feature of the REST API would you use to upload documents into a search index? -Index. -Indexer. -Skillset.

Index. You use the index REST API focused on documents.

Which of the following facial attributes can the Computer Vision service predict? -Location. -Type of eye-glasses. -Occlusion.

Location. The Computer Vision service predicts location for a detected face.

Which of the following option is the best way to manage your search service costs? -Enable enrichment caching if you're using AI enrichment on blob storage. -Keep your search requests and responses inside the Azure datacenter boundary. -Monitor and set budget alerts for all your search resources.

Monitor and set budget alerts for all your search resources. This option is the most effective way to manage your costs.

When translating speech, in which cases can you can use the Synthesizing event to synthesize the translations and speech? -Only when translating to a single target language. -Only when translating to multiple target languages. -When translating to one or more target languages.

Only when translating to a single target language. You can only use event-based synthesis when translating to a single target language.

Which of the following options is a function you can use in a scoring profile? -Tag. -Volume. -Staleness.

Tag. You can alter scores based on common tag values.

What is the limitation of using the Azure Search linked service as a sink in a copy data task? -You can only upload one document at a time. -The JSON can't contain complex data types like arrays. -You have to define the index in the Azure portal first.

The JSON can't contain complex data types like arrays. At the moment, the linked service only supports a limited number of field types.

What information do you need from your Speech service Azure resource to consume it using the Speech SDK? -The location and one of the keys -The primary and secondary keys -The endpoint and one of the keys

The location and one of the keys The Speech SDK requires the location and a key to connect to the Speech service.

Which of the following options is the only supported endpoint for use with Azure Cognitive Search custom AML skill? -Real-time endpoint. -Web service. -Batch endpoint.

Web service. The AmlSkill can use a web service as an endpoint.

What character do you add after a search term boost the term? -+. -^. -!.

^. ^ used in combination with a numerical value boosts a term.

You plan to use the language detection functionality of the Language cognitive service in a container. Which container image should you deploy? -mcr.microsoft.com/azure-cognitive-services/textanalytics -mcr.microsoft.com/azure-cognitive-services -mcr.microsoft.com/azure-cognitive-services/textanalytics/language

mcr.microsoft.com/azure-cognitive-services/textanalytics/language -You must deploy the image that is specific to language detection.

What function of the Translator service should you use to convert the Chinese word "你好" to the English word "Hello"? -detect -translate -transliterate

translate -Translation converts text from one language to another.

Which is the correct custom skill you need to use to connect to an Azure Machine Learning model? -#Microsoft.Skills.Vision.ImageAnalysisSkill. -#Microsoft.Skills.Custom.WebApiSkill. -#Microsoft.Skills.Custom.AmlSkill.

#Microsoft.Skills.Custom.AmlSkill. This skill is the custom skill you can use to connect directly to an AML endpoint.

You have created a search solution. Users want to be able to enter a partial search expression and have the user interface automatically complete the input. What should you add to the index? -A suggester -A synonym map. -A scoring profile.

A suggester A suggester makes it possible to implement autocomplete and suggestions.

You have implemented a custom skill as an Azure function. You want to include the custom skill in your Azure Cognitive Search indexing process. What should you do? -Add a WebApiSkill to a skillset, referencing the Azure function's URI -Create a JSON document with the input schema for your function, and save it in the folder where the documents to be indexed are stored. -Submit each document to the function, and store the output in a separate data source. Then use the Merge skill to add the results to the index.

Add a WebApiSkill to a skillset, referencing the Azure function's URI To integrate an Azure function custom skill into an indexing process, you must define a skillset containing a WebApiSkill with the URI for the function.

You want to include a sentiment score for each document in an index. What should you do? -Create a custom skill that uses an Azure Machine Learning model to predict the sentiment for a document -Create a custom skill that calls the Text Analytics cognitive service to predict the sentiment of each document. -Add the built-in Sentiment skill to the skillset used by the indexer.

Add the built-in Sentiment skill to the skillset used by the indexer. The built-in sentiment skill can be used to accomplish the goal in this scenario.

How should you collect telemetry for your Azure Cognitive Services resource for later analysis? -Create an alert. -Configure diagnostic settings. -Create a dashboard.

Configure diagnostic settings. -Diagnostic settings enable you to capture data for subsequent analysis.

How can you enable users to use your knowledge base though email? -Add Friendly Chit-chat to the knowledge base. -Enable Active Learning for the knowledge base and include the user's email address as the userId parameter in responses. -Create a bot based on your knowledge base and configure an email channel.

Create a bot based on your knowledge base and configure an email channel. You can create a bot for your published knowledge base and configure a channel for email communication.

You want to create a knowledge base from an existing FAQ document. What should you do? -Create an empty knowledge base and manually enter the FAQ questions and answers. -Create a new knowledge base, importing the existing FAQ document. -Create a new knowledge base, selecting only the Professional chit-chat source.

Create a new knowledge base, importing the existing FAQ document. You can create a knowledge base from an existing document or web page.

You want to use the Computer Vision Analyze Image function to generate an appropriate caption for an image. Which visual feature should you specify? -Tags -Description -Categories

Description To generate a caption, include the Description visual feature in your analysis.

Which of the following parameters must you specify when deploying a Cognitive Services container image? -EULA -ResourceGroup -SubscriptionName

EULA -You must specify a EULA parameter with the value "yes" to explicitly accept the license agreement.

You want Video Analyzer for Media to recognize brands in videos recorded from conference calls. What should you do? -Edit the Brands model to show brands suggested by Bing, and add any new brands you want to detect. -Edit the conference call videos to include a caption of each brand seen on their first appearance. -Embed the Video Analyzer for Media widgets in a custom web site that has all the brand images stored for reference.

Edit the Brands model to show brands suggested by Bing, and add any new brands you want to detect. You can both detect known brands, and well as include new brands you want to detect by providing information about it.

When running code on your computer that connects to Cognitive Services, you receive an error that access is denied due to Virtual Network/Firewall rules. What configuration do you need to set in the Cognitive Services instance? -In the Networking properties, configure Selected Networks and Private Endpoints. -In Networking properties, add your client IP address to the Firewall allowed list. -In Access control, add your Azure Active Directory user account to a role.

In Networking properties, add your client IP address to the Firewall allowed list. -The public IP address for your computer can be allowed access in the Firewall settings.

You want to find information in Microsoft Word documents that are stored in an Azure Storage blob container. What should you do to ensure the files can be accessed by Azure Cognitive Search? -Add a JSON file that defines an Azure Cognitive Search index to the blob container -Enable anonymous access for the blob container -In an Azure Cognitive Services resource, and add a data source that references the container where the files are stored

In an Azure Cognitive Services resource, and add a data source that references the container where the files are stored To search files in a blob container, you should create a data source

You want your search solution to show results in descending order of the file_size field value. What is the simplest way to accomplish this goal? -Create a scoring profile that boosts results based on the file_size field -Make the file_size field facetable, and include a facet parameter that specifies the file_size field in queries. -Make the file_size field sortable, and include an orderby parameter that specifies the file_size field in queries.

Make the file_size field sortable, and include an orderby parameter that specifies the file_size field in queries. Making a field sortable enables you to apply an orderby parameter to sort results by that field.

Which API would be best for this scenario? You need to read a large amount of text with high accuracy. Some of the text is handwritten in English and some of it is printed in multiple languages. -Vision API -Read API -Image Analysis API

Read API The Read API is the best choice for large amounts of text and multiple languages.

After an Azure Cognitive Search service has been created, which three metrics can be viewed in graphs without any other configuration? -Search latency, queries per second, and the percentage of throttled queries. -Count of documents processed, count of skills executed, and the search latency. -Number of errors per indexer, number of warnings per indexer, and the total number of documents indexed.

Search latency, queries per second, and the percentage of throttled queries. These three metrics are graphed on the overview pane.

You want to create a skillset that includes a knowledge store definition. Which type of skill should you use to map the enriched fields extracted by your skillset to the desired structure for the knowledge store data? -Merge -Shaper -Split

Shaper A shaper skill enables you to define a custom document structure for your enriched fields.

Which SDK object should you use to specify the language(s) into which you want speech translated? -SpeechConfig -SpeechTranslationConfig -AudioConfig

SpeechTranslationConfig Specify target languages in the SpeechTranslationConfig object.

Which step enables a cognitive search index to store the enrichments from a Language Studio project? -Update your Azure Cognitive Search solution. -Create a function app. -Train a custom text classification model.

Update your Azure Cognitive Search solution. You need to edit the index, indexer, and custom skillset to store the enrichments.

You want Video Analyzer for Media to analyze a video. What must you do first? -Use the Computer Vision service to extract key frames from the video. -Upload the video to Video Analyzer for Media and index it. -Store the video file in an Azure blob store container.

Upload the video to Video Analyzer for Media and index it. You need to index a video before analyzing it.

What's the best way to improve the performance of an AML skill when enriching documents? -Use more powerful nodes in the Kubernetes inference cluster. -Increase the batch size of documents enriched. -Reduce the complexity of the index properties being enriched.

Use more powerful nodes in the Kubernetes inference cluster. The best way to improve efficiency is to improve the performance of your inference cluster.

Your language model needs to detect an email when present in an utterance. What is the simplest way to extract that email? -Use Regular Expression entities. -Use prebuilt entity components. -Use Learned entity components.

Use prebuilt entity components. When a language model needs to detect a common entity, use prebuilt components to have the Language service automatically detect the entity.

You've scanned a letter into PDF format and need to extract the text it contains. What should you do? -Use the Custom Vision service -Use the Image Analysis API of the Computer Vision service. -Use the Read API of the Computer Vision service.

Use the Read API of the Computer Vision service. The Read API can be used to process PDF formatted files.

You need to verify that the person in a photo taken at hospital reception is the same person in a photo taken at a ward entrance 10 minutes later. What should you do? -Create a People Group and add a person for every hospital visitor with multiple photographs to train a model. -Verify the face in the ward photo by comparing it to the detected face ID from the reception photo. -Compare the Age, head pose, and hair color for the faces in the reception and ward photo's.

Verify the face in the ward photo by comparing it to the detected face ID from the reception photo. The most efficient approach is to compare the two faces using the detected face ID within 24 hours.

What function of the Translator service should you use to convert the Russian word "спасибо" in Cyrillic characters to "spasibo" in Latin characters? -detect -translate -transliterate

transliterate -Transliteration converts text from one script to another.

In which format are message exchanged between a client app and a cognitive services resource when using a REST API? -XML -JSON -HTML

JSON -Client apps submit JSON requests and receive JSON responses.

You have created a data source and an index. What must you create to map the data values in the data source to the fields in the index? -A synonym map -An indexer -A suggester

An indexer Use an indexer to map data to index fields.

Your app must interpret a command to book a flight to a specified city, such as "Book a flight to Paris." How should you model the city element of the command? -As an intent. -As an utterance. -As an entity.

As an entity. The city is an entity to which the intent (booking a flight) should be applied.

Which of the following kinds of Azure resource can you use to host a trained Custom Vision model? -Custom Vision (Training) -Computer Vision -Cognitive Services

Cognitive Services You can publish a trained Custom Vision model to either a Custom Vision (Prediction) resource (which is not an option here) or a Cognitive Services resource.

You need to create a facial recognition solution to identify named employees. Which service should you use? -Computer Vision. -Custom Vision. -Face.

Face. Use the Face service to create a facial recognition solution.

You want to create a knowledge store that contains the images extracted from your indexed documents. What kind of projection should you define? -Object -File -Table

File File projections creates a .jpg file for each image extracted from a document.

You're implementing an activity handler for a message received event. How can your code access the text of the message that was received. -Implement a waterfall dialog. -From the turn context. -From the previous member added activity.

From the turn context. The turn context contains relevant information for the current turn - including message text.

You want to train a model that can categorize an image as "cat" or "dog" based on its subject. What kind of Custom Vision project should you create? -Image classification (multiclass) -Image classification (multilabel) -Object detection

Image classification (multiclass) To train a model that classifies an image using a single tag, use an Image classification (multiclass) project.

What languages can the Read API detect? -Only English at this time -Nine languages, including Chinese, French, German, Italian, Japanese, Korean, Portuguese, and Spanish. -Over 160 different languages.

Nine languages, including Chinese, French, German, Italian, Japanese, Korean, Portuguese, and Spanish. There are more than 160 languages available for printed text via the Read API.

Which of the following best describes the predictions made by a machine learning model? -Absolutely correct values based on conditional logic. -Randomly selected values with an equal chance of selection. -Probabilistic values based on correlations found in training data.

Probabilistic values based on correlations found in training data. -Machine learning models are trained using historic data, and rely on algorithms that find statistical relationships in the data. Predictions are generally based on probability; and while models are often extremely accurate, predictions are based on a confidence score that indicates a level of probability.

You want to filter reviews in your app to know how many people like or dislike a business. What feature would you use? -Sentiment analysis -Key phrase extraction -Entity recognition

Sentiment analysis Use sentiment analysis to determine if a natural language query is positive or negative.

How are client applications typically granted access to a cognitive services endpoint? -The application must specify a valid subscription key for the Azure resource. -The user of the application must enter a user name and password associated with the Azure subscription. -Access to cognitive services is granted to anonymous users by default.

The application must specify a valid subscription key for the Azure resource. -By default, access to a cognitive services resource is based on a subscription key.

You plan to use a cognitive services container in a local Docker host. Which of the following statements is true? -Client applications must pass a subscription key to the Azure resource endpoint before using the container. -The container must be able to connect to the Azure resource endpoint to send usage data for billing. -All data passed from the client application to the container is forwarded to the Azure resource endpoint.

The container must be able to connect to the Azure resource endpoint to send usage data for billing. -container usage metrics are sent to the Cognitive Services resource in Azure to calculate billing.

Which SDK object should you use as a proxy for the Translation API of Speech service? -TranslationRecognizer -SpeechRecognizer -SpeechSynthesizer

TranslationRecognizer Use a TranslationRecognizer to call the Translation API of the Speech service.

A data scientist has used Azure Machine Learning to train a machine learning model. How can you use the model in your application? -Use Azure Machine Learning to publish the model as a web service. -Export the model as a cognitive service. -You must build your application using the Azure Machine Learning designer.

Use Azure Machine Learning to publish the model as a web service. -You can use Azure Machine Learning to publish a trained model as a web service, and consume it from applications through its REST interface.

A person needs to build an application that submits expense claims and extracts the merchant, date, and total from scanned receipts. What's the best way to do this? -Use the Read API of the Computer Vision service. -Use Form Recognizer's prebuilt receipts model -Use Form Recognizer's Layout service

Use Form Recognizer's prebuilt receipts model Use the Form Recognizer's prebuilt receipts model. It can intelligently extract the required fields even if the scanned receipts have different names in them.

Where do you copy the full endpoint of the function app to use in the custom web skill that includes the api-version and path to the actual function name? -You can find the endpoint in the Azure portal. -You can find the endpoint in Language Studio. -You can find the endpoint in the Azure extension inside VS Code.

You can find the endpoint in the Azure extension inside VS Code. You right click on the function name in the Azure extension and select Copy function URL.

A person plans to use a Form Recognizer prebuilt invoice model. To extract form data using the model, what are two calls they need to make to the REST API? -Train Model and Get Model Labels -Analyze Invoice and Get Analyze Invoice Result -Create Form Recognizer and Get Analyze Invoice Result

Analyze Invoice and Get Analyze Invoice Result The Analyze Invoice function starts the form analysis and returns a result ID, which they can pass in a subsequent call to the Get Analyze Invoice Result function to retrieve the results.

Which of the following features of Language Studio can you use out of box without needing to train a model? -Conversational language understanding. -Analyze sentiment. -Custom text classification.

Analyze sentiment. The analyze sentiment feature has a pre-built model that you can use without any training.

Which object should you use to specify that the speech input to be transcribed to text is in an audio file? -SpeechConfig -AudioConfig -SpeechRecognizer

AudioConfig Use an AudioConfig to specify the input source for speech.

You want to index a collection of text documents, and search them from a mobile application. Which service should you use to create the index? -The Language cognitive service -Azure Cognitive Search -Azure Bot Service

Azure Cognitive Search -Use Azure Cognitive Search to index documents for search.

What Azure product can you use to enrich an index with different language translations? -Azure Cognitive Search. -Azure Speech Service. -Azure Cognitive Services.

Azure Cognitive Services. Cognitive Services provides translation services.

You want to store the subscription keys for a Cognitive Services resource securely, so that authorized apps can retrieve them when needed. What kind of Azure resource should you provision? -Azure Storage -Azure Key Vault -Azure App Service

Azure Key Vault -Use Azure Key Vault to store credentials securely.

You want to enable a conversational customer support solution through both email and web chat. What should you do? -Create a bot for web chat. Send an automated response to email, directing users to web chat. -Create a bot for email, and a second bot for web chat. -Create a single bot and deliver it through both web chat and email channels.

Create a single bot and deliver it through both web chat and email channels. You can create a single bot and connect it to multiple channels, including web chat and email.

An organization wants to improve the reliability of a search service. It's important that both read and write operations are 99.9% available. Which of these architectures would ensure this reliability? -Create an Azure Cognitive Search service with a Storage Optimized service tier and at least two replicas. -Create an Azure Cognitive Search service with any Standard service tier and at least three replicas. -Create an Azure Cognitive Search service with a High-density service tier and one replica.

Create an Azure Cognitive Search service with any Standard service tier and at least three replicas. The main factor is that the search service has three replicas.

You want to create a search solution that uses a built-in AI skill to determine the language in which each indexed document is written, and enrich the index with a field indicating the language. Which kind of Azure Cognitive Search object must you create? -Synonym map -Skillset -Scoring Profile

Skillset A skillset enables you to define an enrichment pipeline composed of AI skills.

You are defining an alert that notifies you when a key regeneration event is recorded in the activity log for your Cognitive Services resource? What should you do? -Specify a Scope of Activity Log. -Specify an Action that uses an Azure Logic App to read the activity log. -Specify a Condition with a Signal Type of Activity Log.

Specify a Condition with a Signal Type of Activity Log. -The Regenerate Key event is an Activity Log signal.

You need to regenerate the primary subscription key for a Cognitive Services resource that an app uses. What should you do first to minimize service interruption for the app? -Switch the app to use the secondary key -Change the resource endpoint -Enable a firewall

Switch the app to use the secondary key -The app can use the secondary key to access the resource.

You want to create a knowledge store that contains a relational schema for your enriched data. What kind of projection should you define? -Object -File -Table

Table Table projections define a relational schema of tables for your enriched data.

What must you do before taking advantage of the smart labeler tool when creating an object detection model? -Create a JSON file containing bounding box coordinates. -Tag some images with objects of each class and train an initial object detection model. -Train an image classification (multilabel) model.

Tag some images with objects of each class and train an initial object detection model. To take advantage of the smart labeler, tag some images and train an initial model. Subsequently, the portal will suggest tags for new images.

What does an object detection model predict? -The location and class of specific classes of object in an image. -The class of the main subject of an image. -The file type of an image.

The location and class of specific classes of object in an image. Object detection is used to identify bounding boxes containing specific classes of object in an image.

You have analyzed text that contains the word "Paris". How might you determine of this word refers to the French city or the character in Homer's "The Iliad"? -Use the Language service to extract key phrases. -Use the Language service to detect the language of the text. -Use the Language service to extract linked entities.

Use the Language service to extract linked entities. -Linked entities enable you to disambiguate common entities of the same name.

You want to get only the most likely intent for your natural language query. Which parameter should you use? -kind -query -topIntent

topIntent The prediction result topIntent will contain the most likely intent for your query.

A person is building a custom model with Form Recognizer services. What is required to train a model? -Along with the form to analyze, JSON files need to be provided. -Training must be done through language specific SDKs. -Nothing else is required.

Along with the form to analyze, JSON files need to be provided. The labels needed in training are referenced in the ocr.json files, labels.json files, and single fields.json file.

How can you change the voice used in speech synthesis? -Specify a SpeechSynthesisOutputFormat enumeration in the SpeechConfig object. -Set the SpeechSynthesisVoiceName property of the SpeechConfig object to the desired voice name. -Specify a filename in the AudioConfig object.

Set the SpeechSynthesisVoiceName property of the SpeechConfig object to the desired voice name. To set a voice, set the SpeechSynthesisVoiceName property of the SpeechConfig to a voice name, such as "en-GB-George".