Word clouds for text analysis

Word clouds for text analysis

The ‘word cloud’ has become very popular, used increasingly by national newspapers and described as ‘the gateway drug to textual analysis‘ by the Chronicle of Higher Education. Word clouds are now available in agileBase.

Analysing numerical data is fairly well understood. There are all sorts of charts and statistics you can use, many of which have been available in agileBase for some time. However, what do you do when you have databases of unstructured text? For example

• learner records
• book reviews
• customer feedback

How do you discern patterns and pick out trends in this mass of text?

Let’s jump straight in with an example.

This is a summary of the learning database that we use in our organisation – books, online articles, videos, PDFs etc. The text includes our reviews and where relevant summaries from amazon.com. You can quickly get an idea of the sorts of things we’re interested in.

What’s more, the word cloud is interactive – you can click on a word to filter by it. This will show a list of results – books etc. – that you can drill down into and also a new word cloud summarising that set of results. In this way, you can iteratively explore an area and find interesting resources – this is a method we’re pioneering.

Here’s the cloud that you get when you click on ‘systems’:

As you can see, it’s cleverly picked up a number of different forms of the word ‘systems’. This is made possible by the use of research from Lancaster University in natural language processing.

In fact, if you hover over a word, you can see the different forms. For example, science includes also sciencific (a spelling mistake), scientific, scientifically, scientism, scientist and scientists. In this way, you don’t have to know the exact word you’re searching for.

To see a word cloud of your own data, just go to the charts tab in agileBase and select a field to use as a source of text – it’s as simple as that!

We predict this will become a useful resource in many different areas. Let us know what you think, especially if you have an idea that you’d like to try it out on.