Building a Text Summarization API with Azure AI and .NET Core

In a world flooded with information, summarization is no longer a luxury — it’s a necessity. Whether you’re building accessibility tools, content previews, or simplifying dense documents, Azure AI’s Text Analytics service offers a powerful way to distill long-form text into concise summaries.

In this article, we’ll walk through how to build a .NET Core Web API that uses Azure’s Extractive Summarization feature to convert verbose input into short, readable output. We’ll cover:

1.  What to set up in Azure

2.  How the .NET Core wrapper API works

3.  How to test it using Swagger or Postman

 Step 1: Azure Setup — Provision the Language Service

Before writing any code, you need to provision the Azure AI Language Service, which powers the summarization engine.

 What to do:

• Go to the Azure Portal

• Create a new Azure AI Language resource

• Choose a supported region like East US or Sweden Central (required for preview features)

• Select the Standard S pricing tier

• After deployment, copy:

• Endpoint URL

• API Key

These credentials will be used to authenticate your API calls.

🧑‍💻 Step 2: .NET Core API — Wrapping Azure Summarization

Below is the code snippet:

[HttpPost("summarize")]
public async Task<IActionResult> SummarizeText([FromBody] SummarizationRequest request)
{
    if (string.IsNullOrWhiteSpace(request.Text))
        return BadRequest("Text is required.");

    var result = await SummarizeTextAsync(request.Text);

    var allSentences = new List<string>();

    await foreach (var docGroup in result.Value)
    {
        foreach (var doc in docGroup)
        {
            foreach (var sentence in doc.Sentences)
            {
                allSentences.Add(sentence.Text);
            }
        }
    }

    var summary = string.Join(" ", allSentences);
    return Ok(new { summary });



}

private async Task<Azure.AI.TextAnalytics.ExtractiveSummarizeOperation> SummarizeTextAsync(string inputText)
{
    var endpoint = new Uri(AzureAIConstants.LanguageEndpoint);
    var credential = new AzureKeyCredential(AzureAIConstants.LanguageApiKey);
    var _client = new TextAnalyticsClient(endpoint, credential);

    var options = new ExtractiveSummarizeOptions
    {
        MaxSentenceCount = 3,
        OrderBy = ExtractiveSummarySentencesOrder.Rank
    };

    var document = new TextDocumentInput("1", inputText)
    {
        Language = "en"
    };

    var response = await _client.ExtractiveSummarizeAsync(
        WaitUntil.Completed,
        new[] { document },
        options
    );

    return response;



}

public class SummarizationRequest
{
    public string Text { get; set; }
}

Controller Endpoint:

[HttpPost("summarize")]
public async Task<IActionResult> SummarizeText([FromBody] SummarizationRequest request)

This endpoint accepts a POST request with a JSON body containing Text the to summarize. It validates the input and calls the SummarizeTextAsync method.

Azure Client Setup

var endpoint = new Uri(AzureAIConstants.LanguageEndpoint);
var credential = new AzureKeyCredential(AzureAIConstants.LanguageApiKey);
var _client = new TextAnalyticsClient(endpoint, credential);

Here, the Azure Text Analytics client is initialized using the endpoint and key from your Azure resource.

Summarization Options : This configures the summarization behavior:
• MaxSentenceCount: Limits the number of sentences in the summary
• OrderBy: Ranks sentences by relevance

Document Input

var document = new TextDocumentInput("1", inputText)
{
    Language = "en"
};

Azure expects a TextDocumentInput object with a unique ID and language code ( for English).

API Call to Azure

var response = await _client.ExtractiveSummarizeAsync(
    WaitUntil.Completed,
    new[] { document },
    options
);

This sends the document to Azure and waits for the summarization operation to complete.

Extracting Summary Sentences

var allSentences = new List<string>();

await foreach (var docGroup in result.Value)
{
    foreach (var doc in docGroup)
    {
        foreach (var sentence in doc.Sentences)
        {
            allSentences.Add(sentence.Text);
        }
    }
}

var summary = string.Join(" ", allSentences);
return Ok(new { summary });

This loops through the nested response structure and collects the summary sentences into a single string.

Step 3: Testing the API with Swagger or Postman

Once your API is running locally or deployed, you can test it using:

Swagger UI

If you’ve enabled Swagger in your project, navigate to:

http://localhost:<port>/swagger

Find the POST/summarize endpoint, enter a long text, and hit “Execute”. You’ll see the summarized output in the response.

Response:

Final Thoughts

This API is a great starting point for building intelligent content workflows. You can plug it into transcription pipelines, accessibility tools, or even voice narration systems using Azure Speech. With just a few lines of code, you’ve unlocked the power of summarization — and made your content more digestible for everyone.