Quickstart: Query V2 prediction endpoint with C# .NET SDK

Use the .NET SDK, found on NuGet, to send a user utterance to Language Understanding (LUIS) and receive a prediction of the user's intention.

This quickstart sends a user utterance, such as turn on the bedroom light, to a public Language Understanding application, then receives the prediction and displays the top-scoring intent HomeAutomation.TurnOn and the entity HomeAutomation.Room found within the utterance.



The complete solution is available from the cognitive-services-language-understanding GitHub repository.

Looking for more documentation?

Get Cognitive Services or Language Understanding key

In order to use the public app for home automation, you need a valid key for endpoint predictions. You can use either a Cognitive Services key (created below with the Azure CLI), which is valid for many cognitive services, or a Language Understanding key.

Use the following Azure CLI command to create a Cognitive Service key:

az cognitiveservices account create \
    -n my-cog-serv-resource \
    -g my-cog-serv-resource-group \
    --kind CognitiveServices \
    --sku S0 \
    -l WestEurope \ 

Create .NET Core project

Create a .NET Core console project in Visual Studio Community 2017.

  1. Open Visual Studio Community 2017.
  2. Create a new project, from the Visual C# section, choose Console App (.NET Core).
  3. Enter the project name QueryPrediction, leave the remaining default values, and select OK. This creates a simple project with the primary code file named Program.cs.

Add SDK with NuGet

  1. In the Solution Explorer, select the Project in the tree view named QueryPrediction, then right-click. From the menu, select Manage NuGet Packages....
  2. Select Browse then enter Microsoft.Azure.CognitiveServices.Language.LUIS.Runtime. When the package information displays, select Install to install the package into the project.
  3. Add the following using statements to the top of Program.cs. Do not remove the existing using statement for System.
using System.Threading;
using System.Threading.Tasks;
using Microsoft.Azure.CognitiveServices.Language.LUIS.Runtime;
using Microsoft.Azure.CognitiveServices.Language.LUIS.Runtime.Models;

Create a new method for the prediction

Create a new method, GetPrediction to send the query to the query prediction endpoint. The method will create and configure all necessary objects then return a Task with the LuisResult prediction results.

static async  Task<LuisResult> GetPrediction() {

Create credentials object

Add the following code to the GetPrediction method to create the client credentials with your Cognitive Service key.

Replace <REPLACE-WITH-YOUR-KEY> with your Cognitive Service key's region. The key is in the Azure portal on the Keys page for that resource.

// Use Language Understanding or Cognitive Services key
// to create authentication credentials
var endpointPredictionkey = "<REPLACE-WITH-YOUR-KEY>";
var credentials = new ApiKeyServiceClientCredentials(endpointPredictionkey);

Create Language Understanding client

In the GetPrediction method, after the preceding code, add the following code to use the new credentials, creating a LUISRuntimeClient client object.

Replace <REPLACE-WITH-YOUR-KEY-REGION> with your key's region, such as westus. The key region is in the Azure portal on the Overview page for that resource.

// Create Luis client and set endpoint
// region of endpoint must match key's region, for example `westus`
var luisClient = new LUISRuntimeClient(credentials, new System.Net.Http.DelegatingHandler[] { });
luisClient.Endpoint = "https://<REPLACE-WITH-YOUR-KEY-REGION>.api.cognitive.microsoft.com";

Set query parameters

In the GetPrediction method, after the preceding code, add the following code to set the query parameters.

// public Language Understanding Home Automation app
var appId = "df67dcdb-c37d-46af-88e1-8b97951ca1c2";

// query specific to home automation app
var query = "turn on the bedroom light";

// common settings for remaining parameters
Double? timezoneOffset = null;
var verbose = true;
var staging = false;
var spellCheck = false;
String bingSpellCheckKey = null;
var log = false;

Query prediction endpoint

In the GetPrediction method, after the preceding code, add the following code to set the query parameters:

// Create prediction client
var prediction = new Prediction(luisClient);

// get prediction
return await prediction.ResolveAsync(appId, query, timezoneOffset, verbose, staging, spellCheck, bingSpellCheckKey, log, CancellationToken.None);

Display prediction results

Change the Main method to call the new GetPrediction method and return the prediction results:

static void Main(string[] args)

    var luisResult = GetPrediction().Result;

    // Display query
    Console.WriteLine("Query:'{0}'", luisResult.Query);

    // Display most common properties of query result
    Console.WriteLine("Top intent is '{0}' with score {1}", luisResult.TopScoringIntent.Intent,luisResult.TopScoringIntent.Score);

    // Display all entities detected in query utterance
    foreach (var entity in luisResult.Entities)
        Console.WriteLine("{0}:'{1}' begins at position {2} and ends at position {3}", entity.Type, entity.Entity, entity.StartIndex, entity.EndIndex);



Run the project

Build the project in Studio and run the project to the see the output of the query:

Query:'turn on the bedroom light'
Top intent is 'HomeAutomation.TurnOn' with score 0.809439957
HomeAutomation.Room:'bedroom' begins at position 12 and ends at position 18

Next steps

Learn more about the .NET SDK and the .NET reference documentation.