take a walk through this workbook and see what you think. it isn't a straight forward solution, it took some data massaging.
can tell us more about what such a feature would mean to you?
how you would like it to function?
what sort of data exploration capability you envision for such a feature?
I am also interested in seeing this feature being built into Tableau in the future. Recently I have been using SPSS Text Analytics software to do verbatim comments analysis. It allows me to feed in a data file with our organisation's hierarchy and pull out the common themes/messages/categories throughout the verbatims.
I'm also extremely interested in this possibility. I work for a large university, and manage the survey systems for quality assurance and quality enhancement. We've recently moved our quantitative data reporting into Tableau, but this leaves a vast amount of open text qualitative data to be explored:I'm currently leading a project to explore ways to improve our current methods which include using SAS, Excel, Access, Crawdad and Textanalyst with something more streamlined and interactive and you can probably see why I'm keen that Tableau could be an option!
You asked what this would mean to me and how it would function. I'll answer the functionality first, and then the potential impact.
My ideal functionality would be the ability to import free text data in the same way we import closed question data from a survey: linked to demographics we know about a student. Could part of the import process be to identify frequently used key-words, and derive categories based on these? Some form of semantic resonance analysis looking at relationships between key words could also be possible. Once summarised, these key word categories could become dimension in Tableau, with their counts as measures and the keywords themselves as labels. It may then be possible to draw the word in sizes relative to their frequency, and aligned according to the relationships between them, reproducing the functionality of a Wordle map. As the text would have been brought into Tableau, clicking on each of the words in the map could link back to the original full comment so it would be possible to read the context around that original word.
(Don't I make it sound easy?!)
It would mean that all our reporting could be provided in the same system, and that a lot of laborious hand-coding of text could be replaced by exploring the summarised text output using the survey and demographic data. We're keen to get the real insights behind the survey results out to the rest of the university as quickly as possible; Tableau is saving us time with the quantitative data, but the volume of qual information makes this hard to achieve. Being able to speedily move from raw data to insight in this area would be wonderful, and allow real changes to be made to the teaching and services the university delivers.
I'd be happy to talk more about this, if it's a realistic possibility...
Awesome comments, I also have survey data that I would like to look at word by word. It is not exact, but it does provide context.