Hi,
We have a question concerning visualisation of Text analytics.
Within Proximus, there is decided to create one big lexicon of topics and parent topics. Because we have more then 100 topics, and not all are relevant within each survey I have troubles to visualise it in dashboards.
Here are some practical questions.
- How can we exclude topics we don’t want to follow? Or keep only a select set of topics?
If you select on top of the page a certain topic (or for example 10 topics), you will also see all other topics which were in combination with these topics. Because it is a multiple choice with multiple answers possible.
- In our evolution-charts and simple charts => because of > 100 categories, this is not clear. We could hide (uncheck) some topics => but this is a lot of manual work in case of lots of categories. Also does the “sample size” in the widget does not change. In stead of 1000 responses we have only 500 responses for example, but we do not see that. For me this is not a good solution.
- We could add response count threshold => but this can be different by type of widget. Sometimes 100 is enough, sometimes 200 is enough. Or when we have a too broad category where we are not intested in, in this survey, this will make the graph unreadable.
- My request here is: Can you give me a possibility to exclude topics and follow only topics of interest? This would be handy for colleagues that use the dashboard.
- Is it possible to make a Text iQ Table with a topic that was created on Survey level?
- I have the impression NOT and I always need to do text iq on Dashboard level to make this type of chart with topics.