Labelf AI

With Labelf you can create custom classification models that categorize you support interactions automatically. In this guide we will go through how to set up the model.

Click create model in the classification section

Here you pick the relevant data that you wish to categorize

Filters are used to limit the amount of data the classification model can access to categorize. 

Dataset filter ​ Usually you might want to filter the model to only use a certain interaction type or a specific queue or departement. These type of metadata are used in the dataset filter section.

Model filters allow you to create a model that is filtered from another classification model. This means you can create model hierachies. ​ ​

​ ​Generative prompt filter ​ ​This type of filter is used to have your classification be built on a generative model that you have already created and deployed (more on generative models here)

If you want to create a new model from scratch and manually train it, you can press skip in this step and ignore below steps 4.a and 4.b

1. Select label column 
 In this step you can choose to utilize some already set categories to train the model on. This can be used in cases where you already have a manual categorisation that has been done in your customer support system or CX plattform. If choose a column with pre-existing categories the AI model you are creating will use these to start training straight away. Below we see how it would look if we choose the column issue_category.

2. Preview labels
 In the next step we now get a preview of the volume of each category from the chosen column. You then click the categories you wish the model use. This is a good chance to exclude categories that are not relevant, maybe you have some older category in the data that you no longer wish to use. There will also be a warning if a category has less than 100 interactions represented for a certain category.

1. Click create model in the classification section

2. Select the dataset
 
 Here you pick the relevant data that you wish to categorize
 
 
 
3. Add a filter (optional)
 Filters are used to limit the amount of data the classification model can access to categorize. 
 
 
 ​
 
 ​
 Dataset filter ​ Usually you might want to filter the model to only use a certain interaction type or a specific queue or departement. These type of metadata are used in the dataset filter section.
 
 Model filter ​
 Model filters allow you to create a model that is filtered from another classification model. This means you can create model hierachies. ​ ​
 
 ​ ​Generative prompt filter ​ ​This type of filter is used to have your classification be built on a generative model that you have already created and deployed (more on generative models here)
 
4. Use existing categories (optional)
 If you want to create a new model from scratch and manually train it, you can press skip in this step and ignore below steps 4.a and 4.b
 1. Select label column 
 In this step you can choose to utilize some already set categories to train the model on. This can be used in cases where you already have a manual categorisation that has been done in your customer support system or CX plattform. If choose a column with pre-existing categories the AI model you are creating will use these to start training straight away. Below we see how it would look if we choose the column issue_category.
 
 2. Preview labels
 In the next step we now get a preview of the volume of each category from the chosen column. You then click the categories you wish the model use. This is a good chance to exclude categories that are not relevant, maybe you have some older category in the data that you no longer wish to use. There will also be a warning if a category has less than 100 interactions represented for a certain category.
 

5. Model is now created

Your classification model is now created and you can view it in the classification tab.