New Scoring Algorithms

We've been busy at Word Cloud Plus and delivered two updates to our scoring algorithms. Our default algorithm, Combination Counting has been upgraded to improve grouping of long phrases and score reallocation. What does that mean? Using one of our old favourite datasets of 1000 reviews of Disneyland Hong Kong.

Combination Counting (new generation)

Combination Counting (classic)

Getting more from the Data

Our new scoring algorithm already does a better job of pulling out longer phrases like "hong kong disneyland" and "iron man experience" while also giving more prominence to "disneyland". Using more of Word Cloud Plus's features allows us to get even more from the data.

Replace

Sometimes datasets contain common misspellings like MacDonald's instead of McDonald's when discussing the fast-food franchise. Sometimes reviewers forget the apostrophe too, or even call it "Maccy Ds". That is why we created Replace, which allows users to specify text they would like to find and replace. Using Replace, for our Disneyland Hong Kong data we are able to group "hk disneyland" and "hong kong disneyland" so they are analysed together. We can also go as far as treating "disneyland hong kong" as the same as putting "hong kong" first.

After applying our replacements, the phrase "hong kong disneyland" becomes easily the biggest. Therefore, we changed the styling to use the phrase to frame the rest of the word cloud. By grouping some terms, we create space for other phrases to appear in our top 50. By default, all new word clouds use 50 phrases, but this is also configurable.

Coding

If you're a Word Cloud Plus premium subscriber, there is one more feature you can use to tease out even more analytical power, Coding. Using Coding, we are able to highlight phrases on the word cloud and then trace those phrases back through the original text.

Firstly, we have to create codes on the Coding tab.

Then we can return to the word cloud and click on phrases we could like to code.

After applying the coding to the word cloud, we go back to the Coding tab, where we can drill down into the text that drove that phrase up in the scoring.

Finally, after applying codes, we can visit the Text tab and see all of the phrases we have coded have been highlighted throughout the text.

Sign in now to give it a try. Thank you to Kaggle for the Disneyland review data used in these examples.