It is important to understand the difference between topics and categories.
Topics result from unsupervised learning. The topic detection model gives us the sunburst chart (Image below) as a result of hierarchizing up to 3 levels of depth the most relevant topics that appear in our data.
Sunburst Chart
This model lets you dive into your data to understand better what customers, employees, or simply citizens are talking about.
The categories, on the other hand, is a definition that we make based on some examples, where we teach the models what texts they have to group and under what name. Most likely, they will extract example texts from several topics and not from just one.
How to do this:
STEP 1: In the first step, after running a model, the model (Unsupervised) will analyze all the texts and identify which topics appear with greater frequency and relevance among the comments in order to build the SunBurst Chart.
STEP 2: Go to Deepers, upload the dataset there, and build one deeper per each category you want to detect in the data. You can do this by giving positive examples to the deeper and selecting those examples that appropriately represent the category. These categories go across all the topics extracting the comments that are related to some concrete concept, for example, Pricing, Availability, or Shipment problems.
STEP 3: Go back to Analyses, click on the 3 points, and select Detect Category to apply the Deeper to the data and get the category available at the category combo box.
STEP 4: You already will have the category available to filter and check the results for the group of comments that belongs to a specific category.