AzureAI.Community.Microsoft.Semantic.Kernel.PlugIn.Intent 1.0.0-beta6

Azure AI Community GitHub

Intent Detection and Entity Recognition - AzureAI.Community.Microsoft.Semantic.Kernel.PlugIn.Intent

The AzureAI.Community.Microsoft.Semantic.Kernel.PlugIn.Intent plug-in is a powerful tool used to identify and recognize intents and entities within a given text.

Intent Detection

The plug-in allows you to determine the primary intention behind a piece of text. This is valuable in natural language understanding applications, such as chatbots or virtual assistants, where understanding user intentions is crucial.

Entity Recognition

In addition to intent detection, this plug-in excels at entity recognition. It can identify specific entities within the text, providing valuable information that can be used to extract meaningful data.

Whether you're building chatbots, conducting sentiment analysis, or performing any other text-based analysis, the AzureAI.Community.Microsoft.Semantic.Kernel.PlugIn.Intent plug-in is a valuable asset for enhancing the understanding of textual data.

Example: Using AzureAI.Community.Microsoft.Semantic.Kernel.PlugIn.Intent Plug-in

In this example, we demonstrate how to use the AzureAI.Community.Microsoft.Semantic.Kernel.PlugIn.Intent plug-in for intent detection and entity recognition.

IDictionary<string, ISKFunction> conversationSummarySkill =
    kernel.ImportSkill(new IntentPlugIn(kernel));

SKContext intentInformation = await kernel.RunAsync("how is weather in chennai at 4PM",
    conversationSummarySkill["ExtractIntentAndEntity"]);

output:

{
  "text": "how is weather in chennai at 4PM",
  "intent": {
    "name": "Weather Inquiry",
    "score": 0.90
  },
  "entities": [
    {
      "entity": "Location",
      "type": "Location",
      "startPos": 15,
      "endPos": 22,
      "value": "Chennai"
    },
    {
      "entity": "Time",
      "type": "Time",
      "startPos": 25,
      "endPos": 29,
      "value": "4PM"
    }
  ]
}

Customizable Template Examples with the AzureAI.Community.Microsoft.Semantic.Kernel.PlugIn.Intent Plug-in

With the AzureAI.Community.Microsoft.Semantic.Kernel.PlugIn.Intent plug-in, developers have the flexibility to tailor their examples to meet their specific needs, as the default template has been replaced. Furthermore, kindly provide the CoreTheme, and if applicable, an optional list of Intents.

Code Example

You can pass custom examples to the plug-in as follows:

var template1 = new TemplateExamples
{
  "input": "I need to Book my ticket for the 3:30 PM show of 'Learn Microsoft AI - YouTube Channel' at your theater.",
  "output": {
    "text": "I need to Book my ticket for the 3:30 PM show of 'Learn Microsoft AI - YouTube Channel.",
    "intent": {
      "name": "Movie Booking",
      "score": 0.95
    },
    "entities": [
      {
        "entity": "Learn Microsoft AI - YouTube Channel",
        "type": "Movie",
        "startPos": 51,
        "endPos": 63,
        "value": 0,
        "size": null
      },
      {
        "entity": "3:30 PM",
        "type": "Showtime",
        "startPos": 41,
        "endPos": 47,
        "value": 0,
        "size": null
      }
    ]
  }
};

Constructing the Object with Template Information along with Core theme and Intent's list

you need to provide the template information while constructing the object. Here's an example of how to do it:

Example Code

List<string> intentList = new List<string>
{
    "Movie Booking",
    "Rebook",
    "Cancel Booking",
    "Ticket Availability"
};

IDictionary<string, ISKFunction> moviePlugIn = kernel.ImportFunctions(new IntentPlugIn(kernel, "Booking a Movie Ticket", intentList, template1));

unknown intent

If the Core theme or Intent List does not match, the trigger for "unknown intent" is activated.