Skip to content

Latest commit

 

History

History
97 lines (61 loc) · 2 KB

README.md

File metadata and controls

97 lines (61 loc) · 2 KB
Raw

iot-edge-mlnet-text-classification-test

Demonstration of integrating the ML.Net Text Classification into Azure IoT Edge.

This example is part of a blog at https://blog.vandeveldne-online.com

Modules

This project contains two modules:

  1. iot-edge-mlnet-text-classification
  2. iot-edge-mlnet-text-classification-test

The first module takes a message on input 'input1' and write the outcome to output 'output1'. The outcome is also shown in the console log.

The second module is there to test the text classification. It supports a method for sending multiple test comments.

Module iot-edge-mlnet-text-classification

Input message format:

public class Request
{
    public string comment {get; set;}
}

Output message format:

public class Response
{
    public string comment {get; set;}
    public string prediction {get; set;}

    public Score[] scores {get; set;}
}

public class Score
{
    public float entry {get; set;}
}

Module iot-edge-mlnet-text-classification-test

Direct method 'meassureSentiment' body format:

public class Request
{
    public string comment {get; set;}
}

Output message format:

public class Response
{
    public string comment {get; set;}
}

Route

The two modules are conencted with a route:

FROM /messages/modules/test/outputs/output1 INTO BrokeredEndpoint("/modules/text/inputs/input1")

This results in:

DockerHub

The modules are available from Docker hub directly:

svelde/iot-edge-mlnet-text-classification-test:1.0.0
https://hub.docker.com/repository/docker/svelde/iot-edge-mlnet-text-classification-test

and

svelde/iot-edge-mlnet-text-classification:1.0.0
https://hub.docker.com/repository/docker/svelde/iot-edge-mlnet-text-classification

Contribute

This project is open source and availabe under MIT license.

Feel free to contribite to this project.

Or buy us a beer if it realy helped you :-)