Azure AI Language

Sign up for a free Azure account with 200 USD credit and 12 months of popular products, then explore the Azure portal and manage resources for Azure AI language.

Create a new subscription and an Azure AI Language resource in a MAS Language resource group with a storage account; enable sentiment analysis, key phrase extraction, and custom text classification.

Explore how to use the Azure language studio to detect language with the language resource, view confidence scores, and identify English, French, Chinese, and Spanish samples.

Install Visual Studio Code, set up a workspace, and copy endpoint and key from the language resource to connect to the Document Intelligence resource and begin Python or C# development.

Learn to detect language with Python by installing azure-ai-textanalytics, creating a program, authenticating with AzureKeyCredential, and using TextAnalyticsClient to analyze text and return language, iso6391_name, and confidence_score.

Write a C# console app to detect language with Azure Text Analytics, install the Azure AI Text Analytics package, configure endpoint and key, and print language name and confidence score.

Create and configure an Azure AI Language resource for text analytics, including custom text classification and named entity recognition, and implement language detection in Python and C# to test results.

Explore the pre-built answer questions model in language studio, showing how it extracts answers from text across 50 languages, with Surface Laptop-4 examples, short and long answers, and JSON outputs.

Explore the prebuilt Summarise Text model in the Summarise Text tab to generate extractive or abstractive summaries for documents and conversations, supporting 11 languages.

Explore key phrase extraction, which pulls words or phrases from text, supports 94 languages with English default, and presents results in JSON and through a user interface.

Learn to use Azure key phrase extraction in Python by calling client.extract_key_phrases, printing each phrase from response.key_phrases, and loading endpoint and key from a .env file with os.getenv.

Demonstrates using existing C# code to access the key phrase extraction model, convert the response to var, and loop through response.value to extract key phrases via for each.

Discover how the prebuilt named entity recognition (NER) model identifies and categorizes entities, including people, organizations, locations, dates, and more, and supports indexing, search, and automated analysis across 70+ languages.

Learn to use Python to call client.recognize_entities for named entity recognition, loop through response.entities, and extract text, category, subcategory, confidence_score, length, and offset.

Integrate C# code with named entity recognition using client.RecognizeEntities, loop through entities, and extract text, category, subcategory, confidence score, length, and offset.

Explore how prebuilt Language Studio models detect personally identifiable information (PII) and protected health information (PHI), categorize it, and redact sensitive text using Python and C#, with named entity recognition.

Integrate pii detection into existing python code by replacing recognize_entities with recognize_pii_entities and optionally print the redacted_text for the response; filter entities by confidence_score under 0.9.

Apply pii detection to existing c# code by switching to recognize pii entities, displaying redacted text, and filtering low-confidence results to review uncertain entities.

Explore entity linking and disambiguation in Azure AI Language by analyzing how context differentiates meanings like Mars (planet vs chocolate bar), and interpreting linked entities and confidence scores.

Learn to perform entity linking in Python with recognized_linked_entities, extracting entity name, url, data_source, and match text with confidence_score using Mars and moon examples.

Demonstrates using RecognizeLinkedEntities in a C# workflow to extract name, url, and data source from matches, with Mars bar linked to Wikipedia at 86% confidence.

Explore sentiment analysis and opinion mining in Azure AI Language, classifying as positive, neutral, or negative across over 90 languages, with targets, mixed sentiment, assessments, and $1 per 1,000 records.

Extend Python code to perform sentiment analysis and opinion mining using the Azure language client. Inspect document and sentence sentiment and confidence scores to extract targeted opinions.

Learn to integrate existing C# code with sentiment analysis and enable opinion mining using analyze sentiment batch. Explore configuring responses, document and sentence sentiments, and confidence scores.

Upload documents to a blob storage container, enable blob anonymous access, and create a custom text classification project named azurefoodtravel using English, choosing single-label classification with Azure food travel data.

Label documents by class, assign 80/20 training and testing, use preview auto labelling with GPT via Azure OpenAI, then train the azurefoodtravelmodel and evaluate results.

Evaluate model performance using precision, recall, and F1. Review test set results and the confusion matrix, then deploy the azurefoodtravelmodel to East US and test with sample data.

Learn how to integrate custom text classification in Python using Azure AI, including configuring project and deployment names, performing an asynchronous begin_single_label_classify call, and extracting category and confidence score.

Learn to integrate custom text classification in a C# workflow by using a project name and deployment name, preparing input, and calling SingleLabelClassifyAsync with await to obtain category results.

Enable anonymous blob storage, create the environment container, upload documents, label them for 19th century or environment, and train, deploy, and test a custom text classification model in Language Studio.

Explore building a custom named entity recognition model in Language Studio by uploading documents, labeling user, department, and operating system, then prepare for training, deployment, and testing.

Train a custom named entity recognition model, assess its performance with precision and confusion metrics, deploy the model as helpticketdeployment, and test its predictions on sample tickets.

Connect custom text classification to custom NER in python, process documents as help tickets, extract entities with category, subcategory, length, and offset, and learn single and multi-line commenting.

Adapt existing custom text classification code to use custom named entity recognition in C#, iterating over results and extracting entity category and text, with training data tweaks and inline comments.

Create a carservice custom named entity recognition model by labeling make, model, and type of service, training and deploying it, and testing with deployment results.

Explore custom capabilities in language studio, including custom text analytics for health, sentiment, custom question answer, conversational language understanding, summarization, and translation, plus guidance on deleting resources and exporting projects.

Advance your Azure AI Language skills by deploying resources, using models, and building custom text classification and named entity recognition in Python or C#. Prepare for the applied skills credential.

Review Azure language workflow: create an Azure language resource, analyze documents with prebuilt models, and extract information using Python and C#. Practice, take the exam, earn your credential.