[csharp sdk] TextAnalytics 20251115preview (#54033)

* initiate first version of sdk

* adjust old tests

* add new live tests for new features

* updated samples

* updated record tag

* removed the seperate sln

* updated tag

* revert tsp-location in Conv authoring

* run Export-API.ps1

* updated general readme

* update comit id

* update samples

* updated snippets

* update changelog

Author: amber-Chen-86

sdk/cognitivelanguage/Azure.AI.Language.Text/Azure.AI.Language.Text.sln

@@ -1,14 +1,15 @@
 # Release History
 
-## 1.0.0-beta.4 (Unreleased)
+## 1.0.0-beta.4 (2025-11-20)
 
-### Features Added
-
-### Breaking Changes
+This version of the client library defaults to the service API version `2025-11-15-preview`.
 
-### Bugs Fixed
+### Features Added
 
-### Other Changes
+- Added support for **multiple redaction policies** in a single request.
+- Added **synthetic replacement redaction**, enabling selected PII types (e.g., Person, Email) to be replaced with realistic synthetic values rather than masked.
+- Added **confidence score thresholding**, allowing customers to define minimum confidence levels—globally or per-entity—so that only entities meeting the required confidence are returned.
+- Added **entity validation control** with the `disableEntityValidation` parameter, allowing users to bypass entity validation when needed.
 
 ## 1.0.0-beta.3 (2025-06-23)
 

sdk/cognitivelanguage/Azure.AI.Language.Text/CHANGELOG.md

@@ -17,7 +17,7 @@ Text Analytics is part of the Azure Cognitive Service for Language, a cloud-base
   [Source code][source_root] | [Package (NuGet)][package]| [API reference documentation][text_refdocs] | [Product documentation][text_docs] | [Samples][source_samples]
 
 > [!NOTE]
-> Text Authoring is not supported in version 2.0.0-beta.1. If you use Text Authoring, please continue to use version 1.1.0. You can find the [samples][textauthoring_samples] here.
+> Text Authoring is not supported from version 2.0.0-beta.1. If you use Text Authoring, please use the separate Text Authoring SDK. You can find the [samples][textauthoring_samples] here.
 
 ## Getting started
 
@@ -31,6 +31,7 @@ dotnet add package Azure.AI.Language.Text --prerelease
 
 |SDK version  |Supported API version of service
 |-------------|------------------------------------------------------------------------------------------------
+|1.0.0-beta.4 | 2022-05-01, 2023-04-01, 2024-11-01, 2025-05-15-preview, 2025-11-15-preview (default)
 |1.0.0-beta.3 | 2022-05-01, 2023-04-01, 2024-11-01, 2024-11-15-preview, 2025-05-15-preview (default)
 |1.0.0-beta.2 | 2022-05-01, 2023-04-01, 2024-11-01, 2024-11-15-preview (default)
 |1.0.0-beta.1 | 2022-05-01, 2023-04-01, 2023-11-15-preview (default)

sdk/cognitivelanguage/Azure.AI.Language.Text/README.md

@@ -2,5 +2,5 @@
     "AssetsRepo": "Azure/azure-sdk-assets",
     "AssetsRepoPrefixPath": "net",
     "TagPrefix": "net/cognitivelanguage/Azure.AI.Language.Text",
-    "Tag": "net/cognitivelanguage/Azure.AI.Language.Text_66a454ef6f"
+    "Tag": "net/cognitivelanguage/Azure.AI.Language.Text_48cbf61c49"
 }

sdk/cognitivelanguage/Azure.AI.Language.Text/api/Azure.AI.Language.Text.net8.0.cs

@@ -17,7 +17,7 @@ The values of the `endpoint` and `apiKey` variables can be retrieved from enviro
 
 ## Detect the language of documents
 
-To detect the language of a document, call `AnalyzeText` on the `TextAnalysisClient`, which returns a `AnalyzeTextLanguageDetectionResult` object with the name of the language, a confidence score, and more.
+To detect the language of a document, call `AnalyzeTextAsync` on the `TextAnalysisClient`, which returns a `AnalyzeTextLanguageDetectionResult` object with the name of the language, a confidence score, and more.
 
 ```C# Snippet:Sample1_AnalyzeTextAsync_LanguageDetection
 string textA =
@@ -81,11 +81,11 @@ catch (RequestFailedException exception)
 }
 ```
 
-To detect the language of a document, call `AnalyzeText` on the `TextAnalysisClient`, which returns a `AnalyzeTextLanguageDetectionResult` object with the name of the language, a confidence score, and more.
+To detect the language of a document, call `AnalyzeTextAsync` on the `TextAnalysisClient`, which returns a `AnalyzeTextLanguageDetectionResult` object with the name of the language, a confidence score, and more.
 
 ## Detect the language of documents with country hints
 
-If the country where a document originates from is known, you can aid the language detection model if you call `AnalyzeText` on the `TextAnalysisClient` while passing the documents as an `IEnumerable<LanguageInput>` parameter, having set the `CountryHint` property on each `LanguageInput` object accordingly.
+If the country where a document originates from is known, you can aid the language detection model if you call `AnalyzeTextAsync` on the `TextAnalysisClient` while passing the documents as an `IEnumerable<LanguageInput>` parameter, having set the `CountryHint` property on each `LanguageInput` object accordingly.
 
 ```C# Snippet:Sample1_AnalyzeTextAsync_LanguageDetection_CountryHint
 string textA =

sdk/cognitivelanguage/Azure.AI.Language.Text/api/Azure.AI.Language.Text.netstandard2.0.cs

@@ -17,7 +17,7 @@ The values of the `endpoint` and `apiKey` variables can be retrieved from enviro
 
 ## Extracting key phrases from multiple documents
 
-To extract key phrases from multiple documents, call `AnalyzeText` on an `AnalyzeTextInput`.  The results are returned as a `AnalyzeTextKeyPhraseResult`.
+To extract key phrases from multiple documents, call `AnalyzeTextAsync` on an `AnalyzeTextInput`.  The results are returned as a `AnalyzeTextKeyPhraseResult`.
 
 ```C# Snippet:Sample3_AnalyzeTextAsync_ExtractKeyPhrases
 string textA =

sdk/cognitivelanguage/Azure.AI.Language.Text/assets.json

@@ -17,7 +17,7 @@ The values of the `endpoint` and `apiKey` variables can be retrieved from enviro
 
 ## Recognize entities in multiple documents
 
-To recognize entities in multiple documents, call `AnalyzeText` on the `TextAnalysisClient` by passing the documents as an `AnalyzeTextInput` parameter. This returns a `AnalyzeTextEntitiesResult`.
+To recognize entities in multiple documents, call `AnalyzeTextAsync` on the `TextAnalysisClient` by passing the documents as an `AnalyzeTextInput` parameter. This returns a `AnalyzeTextEntitiesResult`.
 
 ```C# Snippet:Sample4_AnalyzeTextAsync_RecognizeEntities
 string textA =
@@ -102,12 +102,12 @@ For more information on the changes in the preview API version, see [Preview API
 
 To create a new `TextAnalysisClient` with the preview API version, you will need the service endpoint and credentials of your Language resource with the `TextAnalysisClientOptions` pointing to the preview API Version.
 
-To recognize entities in multiple documents, call `AnalyzeText` on the `TextAnalysisClient` by passing the documents and actionContent as an `AnalyzeTextInput` parameter. This returns a `AnalyzeTextEntitiesResult`.
+To recognize entities in multiple documents, call `AnalyzeTextAsync` on the `TextAnalysisClient` by passing the documents and actionContent as an `AnalyzeTextInput` parameter. This returns a `AnalyzeTextEntitiesResult`.
 
 ```C# Snippet:Sample4_AnalyzeTextAsync_RecognizeEntities_Preview
 Uri endpoint = TestEnvironment.Endpoint;
 AzureKeyCredential credential = new(TestEnvironment.ApiKey);
-TextAnalysisClientOptions options = new TextAnalysisClientOptions(TextAnalysisClientOptions.ServiceVersion.V2024_11_15_Preview);
+TextAnalysisClientOptions options = new TextAnalysisClientOptions(TextAnalysisClientOptions.ServiceVersion.V2025_11_15_Preview);
 var client = new TextAnalysisClient(endpoint, credential, options);
 
 string textA =

sdk/cognitivelanguage/Azure.AI.Language.Text/samples/Sample1_AnalyzeTextAsync_LanguageDetection.md

@@ -17,7 +17,7 @@ The values of the `endpoint` and `apiKey` variables can be retrieved from enviro
 
 ## Recognizing Personally Identifiable Information in multiple documents
 
-To recognize Personally Identifiable Information in multiple documents, call `AnalyzeText` on an `TextPiiEntitiesRecognitionInput`.  The results are returned as a `AnalyzeTextPiiResult`.
+To recognize Personally Identifiable Information in multiple documents, call `AnalyzeTextAsync` on an `TextPiiEntitiesRecognitionInput`.  The results are returned as a `AnalyzeTextPiiResult`.
 
 ```C# Snippet:Sample5_AnalyzeTextAsync_RecognizePii
 string textA =
@@ -84,7 +84,7 @@ foreach (DocumentError analyzeTextDocumentError in piiTaskResult.Results.Errors)
 
 ## Recognizing Personally Identifiable Information in multiple documents with a redaction policy
 
-To recognize Personally Identifiable Information in multiple documents, call `AnalyzeText` on an `TextPiiEntitiesRecognitionInput`.  The results are returned as a `AnalyzeTextPiiResult`.
+To recognize Personally Identifiable Information in multiple documents, call `AnalyzeTextAsync` on an `TextPiiEntitiesRecognitionInput`.  The results are returned as a `AnalyzeTextPiiResult`.
 
 ```C# Snippet:Sample5_AnalyzeTextAsync_RecognizePii_RedactionPolicy
 string textA =
@@ -114,7 +114,15 @@ AnalyzeTextInput body = new TextPiiEntitiesRecognitionInput()
     {
         ModelVersion = "latest",
         // Avaliable RedactionPolicies: EntityMaskPolicyType, CharacterMaskPolicyType, and NoMaskPolicyType
-        RedactionPolicy = new EntityMaskPolicyType()
+        RedactionPolicies =
+    {
+    new EntityMaskPolicyType
+    {
+        // defaultPolicy: use entity mask for everything unless overridden
+        PolicyName = "defaultPolicy",
+        IsDefault = true,
+    }
+    }
     }
 };
 
@@ -352,6 +360,157 @@ foreach (DocumentError error in result.Results.Errors)
 }
 ```
 
+## Recognizing Personally Identifiable Information with multiple redaction policies and with synthetic mask
+
+To recognize Personally Identifiable Information with multiple redaction policies, call `AnalyzeTextAsync` on an `TextPiiEntitiesRecognitionInput`.  The results are returned as a `AnalyzeTextPiiResult`.
+
+```C# Snippet:Sample5_AnalyzeTextAsync_RecognizePii_RedactionPolicies
+string documentText = "My name is John Doe. My ssn is 123-45-6789. My email is [email protected]..";
+
+AnalyzeTextInput body = new TextPiiEntitiesRecognitionInput
+{
+    TextInput = new MultiLanguageTextInput
+    {
+        MultiLanguageInputs =
+{
+    new MultiLanguageInput("A", documentText) { Language = "en" },
+    new MultiLanguageInput("B", documentText) { Language = "en" },
+}
+    },
+    ActionContent = new PiiActionContent
+    {
+        PiiCategories = { PiiCategory.All },
+
+        RedactionPolicies =
+{
+    new EntityMaskPolicyType
+    {
+        // defaultPolicy: use entity mask for everything unless overridden
+        PolicyName = "defaultPolicy",
+        IsDefault = true,
+    },
+    new CharacterMaskPolicyType
+    {
+        // customMaskForSSN: keep the last 4 digits of SSN, mask the rest
+        PolicyName = "customMaskForSSN",
+        UnmaskLength = 4,
+        UnmaskFromEnd = false,
+        EntityTypes =
+        {
+            PiiCategoriesExclude.UsSocialSecurityNumber
+        },
+    },
+    new SyntheticReplacementPolicyType
+    {
+        // syntheticMaskForPerson: generate synthetic values for Person and Email
+        PolicyName = "syntheticMaskForPerson",
+        EntityTypes =
+        {
+            PiiCategoriesExclude.Person,
+            PiiCategoriesExclude.Email
+        },
+    }
+}
+    }
+};
+
+Response<AnalyzeTextResult> response = await client.AnalyzeTextAsync(body);
+AnalyzeTextPiiResult piiTaskResult = (AnalyzeTextPiiResult)response.Value;
+
+foreach (PiiActionResult piiResult in piiTaskResult.Results.Documents)
+{
+    Console.WriteLine($"Result for document with Id = \"{piiResult.Id}\":");
+    Console.WriteLine($"  Redacted Text: \"{piiResult.RedactedText}\"");
+    Console.WriteLine($"  Recognized {piiResult.Entities.Count} entities:");
+
+    foreach (PiiEntity entity in piiResult.Entities)
+    {
+        Console.WriteLine($"    Text: {entity.Text}");
+        Console.WriteLine($"    Offset: {entity.Offset}");
+        Console.WriteLine($"    Length: {entity.Length}");
+        Console.WriteLine($"    Category: {entity.Category}");
+        if (!string.IsNullOrEmpty(entity.Subcategory))
+        {
+            Console.WriteLine($"    SubCategory: {entity.Subcategory}");
+        }
+        Console.WriteLine($"    Confidence score: {entity.ConfidenceScore}");
+        Console.WriteLine();
+    }
+
+    Console.WriteLine();
+}
+
+foreach (DocumentError analyzeTextDocumentError in piiTaskResult.Results.Errors)
+{
+    Console.WriteLine($"  Error on document {analyzeTextDocumentError.Id}!");
+    Console.WriteLine($"  Document error code: {analyzeTextDocumentError.Error.Code}");
+    Console.WriteLine($"  Message: {analyzeTextDocumentError.Error.Message}");
+    Console.WriteLine();
+}
+```
+
+## Recognizing Personally Identifiable Information with confidence score threshold
+
+To recognize Personally Identifiable Information with confidence score threshold, call `AnalyzeTextAsync` on an `TextPiiEntitiesRecognitionInput`.  The results are returned as a `AnalyzeTextPiiResult`.
+
+```C# Snippet:Sample5_AnalyzeTextAsync_RecognizePii_ConfidenceScoreThreshold
+    string text =
+        "My name is John Doe. My ssn is 222-45-6789. My email is [email protected]. John Doe is my name.";
+
+    // Input documents
+    MultiLanguageTextInput textInput = new MultiLanguageTextInput
+    {
+        MultiLanguageInputs =
+{
+    new MultiLanguageInput("1", text) { Language = "en" }
+}
+    };
+
+    // Confidence score overrides:
+    //   default = 0.3
+    //   SSN & Email overridden to 0.9 (so they get filtered out as entities)
+    ConfidenceScoreThreshold confidenceThreshold = new ConfidenceScoreThreshold(0.3f);
+    confidenceThreshold.Overrides.Add(
+        new ConfidenceScoreThresholdOverride(
+            value: 0.9f,
+            entity: PiiCategory.UsSocialSecurityNumber.ToString()
+        ));
+    confidenceThreshold.Overrides.Add(
+        new ConfidenceScoreThresholdOverride(
+            value: 0.9f,
+            entity: PiiCategory.Email.ToString()
+        ));
+
+    PiiActionContent actionContent = new PiiActionContent
+    {
+        PiiCategories = { PiiCategory.All },
+        DisableEntityValidation = true,
+        ConfidenceScoreThreshold = confidenceThreshold
+    };
+
+    AnalyzeTextInput body = new TextPiiEntitiesRecognitionInput
+    {
+        TextInput = textInput,
+        ActionContent = actionContent
+    };
+
+    Response<AnalyzeTextResult> response = await client.AnalyzeTextAsync(body);
+    AnalyzeTextPiiResult piiResult = (AnalyzeTextPiiResult)response.Value;
+
+    PiiActionResult doc = piiResult.Results.Documents[0];
+
+    Console.WriteLine($"Redacted text: \"{doc.RedactedText}\"");
+    Console.WriteLine("Recognized entities (after confidence score filtering):");
+
+    foreach (PiiEntity entity in doc.Entities)
+    {
+        Console.WriteLine($"  Text: {entity.Text}");
+        Console.WriteLine($"  Category: {entity.Category}");
+        Console.WriteLine($"  Confidence score: {entity.ConfidenceScore}");
+        Console.WriteLine();
+    }
+```
+
 See the [README] of the Text Analytics client library for more information, including useful links and instructions.
 
 [DefaultAzureCredential]: https://github.com/Azure/azure-sdk-for-net/blob/main/sdk/identity/Azure.Identity/README.md