Collocations are words that are often seen together (like a married couple). In fact, it feels awkward to see them fooling around with another word. An example of a collocation is a phrase like “Powerful Computer”. You could say “Strong Computer” and it wouldn’t be wrong. It just wouldn’t feel right. (Some other examples include “crystal clear”, “strong tea” , “middle management”.)
You can ask Python’s Natural Language Toolkit to return a specified number of Collocations.
From that, an idea for a sample banner ad pops into mind. What if we showed a woman, from behind, wearing a CustomInk-made t-shirt. The headline of the banner is “Quick Turnaround”. And sure enough, we see her animated as she “turns around” and we now see the front of the t-shirt.
Is this the greatest and most Creative banner ever written? No, certainly not. But I hope it helps you make a connection as to how we can take raw text data and mine it (after much preprocessing) and use it as Creative fodder for brainstorming. Certainly, there’s a lot more to mine in these 10,000 Customer Comments.
You can also place the word “turnaround” into an NLTK Concordance method (shown below) and see that, despite this being a collocation, customers have used a number of different adjectives in praising CustomInk’s turnaround times. (Fast, super fast, extra fast, really quick, overall, delighted, 48 hour, short, impressive, great, amazing, speedy, pleased, tight, awesome, friendly and sluggish (uh, oh – one for customer service to explore). We could easily make a cool 15 second, motion graphic video content from that.
In the output below, we can see that some version of “adjective_turnaround” shows up 154 times. During the CustomInk new biz pitch, I seem to recall “fast turnaround” as being one of the bullet points mentioned in the brief. I don’t recall anyone doing a spot or even mentioning it in a spot. If we had this data, I think we might have thought more about presenting it as a concept. I think the client would have been at least a little impressed by this data-based idea.