Running the Genius Task

Set up the text analytics task. To do this:

1. Go to the Survey Management > Text Analytics menu item.

The Text Analytics overlay opens - .



Figure 1 - The Text Analytics overlay

2. Enter the Model ID (you can find this in Model Builder).
3. Select “with categories and attributes” to use the hierarchical model.
4. Select "Synchronize model to database designer tables" to automatically sync your category list from Model Builder. This means that you will not then have to worry about making manual changes if you have made changes.
5. Use "with full path in label" if you have the pipe (|) character in your category list, for use in the Text Analytics template report.
6. Enter the variable ID – this is the name of the open text question that you are analyzing.

Note: Variable IDs are case sensitive, so ensure they are exactly the same as the Qids. Note also that you must run each variable ID separately.

7. Extra Variable IDs – you can bring in background variables (currently single variables are supported) to increase the accuracy of your expressions. Type the name of the survey variable(s) that you want brought into Model Builder for use in your model.
8. You may wish to add in an expression, for example language=="9" (only looking at English verbatim) or interview_start > 10/1/2015 (only processing from a certain date).

Note: For more assistance with building expressions, you can use the expression builder. Go to Survey Data > Export and then click on the Edit link below the expression box. You will see a way to build the expression that you need. Copy that and paste it into the Text Analytics task expression box.

9. Set the start time for the analysis.

You can make the analysis task recurring .

Note: If you have more than 50,000 verbatim to process, set up a recurring task as only the first 50,000 verbatim will be analyzed. When the next task is scheduled to run, it will pick up the remaining verbatim and process those as well.

10. Click Process Variables.

The analysis task will be run according to the Start Time settings.

Note: Only responses in which the text field contains text will be sent for analysis. Responses in which the text field is empty will be ignored.

Note: If you set the analysis to run as a recurring task, the first time it is run it will process all the data while subsequent runs will process only new or changed data.