A categorization model must be built around how your customers are speaking about your business; borrowing a model from another business will not work. So before you start to build the categorization structure you will need to have a fairly detailed understanding of what your customers are saying about you. Based on this you can then identify the major themes or categories that define your business structure, and decide where Text Analytics will provide value by reporting sentiment about each category.
When creating a new model it is best to start simple and to build up more complex categories as you explore your verbatim data. Having identified the main categories, you must then identify the sub-categories. Here you should start with simple sub-categories; you can always create further sub-categories as you find them.
You can then create the expression for each sub-category.
Identify what customers are saying about these categories that you want to be able to measure; these will be your Attributes that the analysis will look for.
Create the expressions for the Attributes.
Warning
To avoid disappointment, save your work at regular intervals and ensure you save before you leave Model Builder unattended (go to Inactive-Screen Lock for more information).
Model Overview
The Overview tab contains all the background information about your model - .
Figure 1 - Example of the Model Overview tab
Use this tab to:
Edit your model name (click on the edit icon next to the model name).
Last Modified - See who created and who last modified your model, with the dates.
Status – shows an orange Design button if the model has not yet been saved and marked as Active (this also means that it cannot be used in ) and a green Active button if the model has been saved as a version. The first time it is saved as a version, it is set to Active. Note that you can only use a model in if it has the green Active button.
Last saved – the date that this model was last saved as a version.
Concept Miner Core – this gives you a list of the available cores, which can be added to your model to help with the analysis.
Sentiment – you can select the sentiment configuration from the drop down menu. If you do not know which configuration to use, speak to your Account Manager.
Uncaptured attributes – an on/off switch. If set to On, uncaptured attributes are generated automatically in your model. If set to off, the Uncaptured attributes folder is not generated.
Get list of categories and attributes list – displays the list of categories, sub-categories, attributes and their associated IDs, along with the parent hierarchy reference, which you need to copy and paste into a database designer table.
Survey variables (feature toggled) – if you have brought any survey variables through in your Text Analytics task, you will see the name, codes and answers in this section (go to Survey Variables - Adding Survey Context into Models for more information)
Versions - a list of saved versions of the model, along with any notes that you have written when saving the version. Click on the name and it will take you into that saved version, so that you can test the model and view the results of the expressions. From this screen you can duplicate this version of the model, which copies that version and gives it a new model ID. The version that is currently active will have a green Active button next to it.
The buttons towards the upper-right corner of the screen are:
Duplicate – makes a copy of the model and gives the duplicate a new model ID. You are redirected to the new model immediately, and the new model is called ‘Copy of ...’
Archive – archiving a model takes it out of your Model list and puts it into your Archived models list. To view the Archived models list, click on the Active models drop-down on the Model List page (go to Archiving Models for more information). From the Archived models list you can restore the model back into your Active models list.
Move Model – if you have access to more than one account you can move a model from one account to the other. Moving the model removes it from the originating account and puts it into the destination account.