Tribal Tool-Kit

Word clouds from Facebook Fan Pages

The Mocks fan page is very active, so I thought it would be interesting to create a word cloud of the comments from: What kind of weekend did you and your Mock have – in one word?

Some comments had a little story, and we still included these.

This is what we did to get the word cloud below:

• Copy and pasted the text into a spreadsheet
• Deleted the profile pic, time and ‘comment’. This left the comments.
• Did a few find & replaces.
  • Took out all the symbols by finding . , ) [ ! etc and replacing with nothing
  • To find spaces and replace with comma and space. This allows CloudMaker to make a series of words into separate words for the word cloud.
• Saved as a CSV file.
• Uploaded the CSV to Tribal Tool-Kit.
• Clicked on the ‘Amalgamate similar terms’ link (this will merge the same words so your words are easier to edit).
• Added a list of words to the stopword list. These were: i; my; to; and; the; a; are; comment; dont; for; im; in; it; of; they; still; is; come; with. This means that these words were still in the list of data, but won’t appear in the word cloud.
• Deleted from the list: don’t
• Merged some words that were similar so that they had a higher frequency and therefore appeared bigger:
  • Supercalifragilisticexpialidocious and supercalifragolisticexpialidociousi
  • BORING and boringi
  • disappointing and disappoint
  • Mockariffic and Mockorific
  • Mocktastic and mocktasstic
• Clicked on ‘Create word cloud from dataset’.
• Changed the font to: Comic Sans MS (Bold Regular).
• Changed ‘Convert case’ to ‘all lower case’
• Made the maximum frequency colour black (#000000)
• Made the minimum frequency colour pink (#CC3399)
• Changed the ‘Save options and formatting’ to ‘new template’
• Clicked on ‘re-draw Word Cloud’
• Gave the template the name ‘Mocks’ and clicked ‘Save’. There is an option to make the settings the default template so future word clouds have this format as soon as you click on the ‘Create word cloud from dataset’.
• Then clicked on ‘Save as image options’. You can save the word cloud as an SVG, PNG, or JPEG image format. JPEG is the lowest quality but opens in the most applications. The word cloud to the right is a JPEG format.

Simple. And interesting. Lipgloss is so big because we kept the 3 times it was said in the one comment shown above. Great to see the number of ways that fans put ‘mock’ into a word and that ‘supercalifragilisticexpialidocious’ was used more than once!

Scenario: You have done a survey and you want to get a quick understanding of the words participants used to answer an open response question.

Solution: A perfect way to do this is to make a word cloud – a visual way to understand the frequency of words; where words with a higher frequency are larger, and words with a lower frequency are smaller.

Problem: The English language has two main spelling systems – the British system and the American system. Read more about the differences at Wikipedia.

Implication: The two spelling systems result in a lower overall frequency for essentially the same word, as they are considered 2 words, and therefore a smaller size in a word cloud.

For example, localise and localize are the same word. If each are used 5 times by participants, the two words would be smaller than if they were combined to have a frequency of 10 using the spelling of your preference.

To show the impact this has on a word cloud, I selected a group of words with different spelling and put them into a spreadsheet. To create a frequency, I used a formula to count the number of characters in the word [In Excel this is LEN(text)].

Most word cloud software only allows you to paste in a group of words or upload a file of words, before generating the cloud. You can sometimes automatically merge similar words (for example when there is the word, the plural, and end with ‘ing’ they will merge to be one word with the combined frequency). I haven’t found one, other than CloudMaker, that allows you to personally merge similar words, enabling you to handle the problem of British and American English.

Below, the first word cloud is all the words and to the second word cloud is the merged list.

Fewer words makes it easier to understand but also changes the priorities.

All the words

Merged words

Impact: When words with British and American spelling are mixed with words spelt the same in both systems, the first impression views could be inaccurate.

For example, if there was a single spelt word, such as: national, with the frequency of 10 and one of the dual spelt words, such as: localise with the frequency of 7, then also localize with a frequency of 5, merging localise and localize results in a frequency of 12, which is greater than the single spelt word, national, with a frequency of 10.

This could change your thinking about how the question was answered as localise is more frequent than national.

If the question was: What should our regional focus be? Then merging the British and American systems would result in a different first view, than looking at a word cloud without merging – because localise would be greater than national rather than the reverse when not merged.