Microsoft Introduces the Next Generation of the Conversational Language Understanding Client Library

The demand for intelligent technologies that can interpret brief text has increased. As increasingly sophisticated solutions are produced, there is a greater need to improve and facilitate the creation of these complex situations. These scenarios are intelligent customer assistance bots to independent computers that interpret human input. The Language Cognitive Service has opted to employ a multilingual transformer-based paradigm to deal with such problems. When using this model, customers will notice a considerable increase in performance over the old Language Understanding Service (LUIS).

Microsoft has released the next generation Conversational Language Understanding client library, allowing developers to use the Azure Cloud Conversational Language Understanding service to train models and use them in applications to provide related language services. Developers can use .NET or Python, and these libraries are currently under beta development.

Language Cognitive Service uses the multi-language Transformer model. According to the official, developers will notice a considerable boost in performance compared to existing language understanding services.

Conversational language understanding is a cloud service that allows developers to train conversation language models using sophisticated technologies. Azure Cognitive Services includes the conversation language understanding API, a collection of artificial intelligence, and cloud machine learning training methods. Advanced technology may be used by developers to train conversational language models and then with a REST API or a client library.

This conversation understanding client library may be used by developers to construct various conversation apps, such as analyzing the dialogue of a chatbot scene. Using text dialogues supplied by the end-user as input, the client is called to execute synchronization activities. For analysis, specify the model and deployment slot. Developers can offer phrases in many languages to employ integrated multilingual functionality for the same model.

The new CLU preview service allows you to construct orchestration projects. The multilingual Dual Transformer Encoder (DTE) powers these orchestration projects. The DTE enables you to route a project to numerous customized language services. These services include knowledge bases that answer questions, other CLU projects, and traditional LUIS apps.

Consider the following example to see how a user may use CLU to analyze a conversational speech in a chatbot situation. As input, the chatbot user would supply a written utterance. This phrase might be a transcription of spoken language. This input phrase would be delivered synchronously to a conversational endpoint with the matching key and project name. These parameters can be saved as an environment variable, a configuration option, or any other method that makes sense for your project.

var endpoint = new Uri("");
var credential = new AzureKeyCredential("{api-key}");

var client = new ConversationAnalysisClient(endpoint, credential);

Let’s look at some code to see how the app may use this library. Although you may use any of our languages (.NET or Python), the examples supplied employ .NET. These libraries are in beta right now and support the new CLU service.

Examine a conversation

You might contact the client as a synchronous operation given a user input phrase to examine the input for a specified model name and deployment slot.

Response<AnalyzeConversationResult> response = client.AnalyzeConversation(
    "We'll have 2 plates of seared salmon nigiri.");

Console.WriteLine($"Top intent: {response.Value.Prediction.TopIntent}");

You may also submit one word in each language to the same model to leverage the inbuilt multilingual capability directly. You might send a phrase in a foreign language to interpret it. Alternatively, while delivering the words to the endpoint, identify the language as a parameter:

Response<AnalyzeConversationResult> response = client.AnalyzeConversation(
    "Tendremos 2 platos de nigiri de salmΓ³n braseado.");
    Language = "es"
Console.WriteLine($"Top intent: {response.Value.Prediction.TopIntent}");

Resources for this library:

For further reading, consider this.


🐝 Join the Fastest Growing AI Research Newsletter Read by Researchers from Google + NVIDIA + Meta + Stanford + MIT + Microsoft and many others...