Insight 101; an image is worth a thousand words:
As vision in humans constitutes the most important of all our senses (IMO),
one quick and pictorial way to develop intuition about the nature
of the reviews for a given rum is to generate what is known as a
word-cloud representation. Such representation is an image that
displays the N-most common words in a body of text after removing those
frequent terms that do not contribute any meaning to it (known as stop-words)
(examples are articles, pronouns, etc). In the cloud image,
common words are displayed such that their
font sizes are made proportional to the word frequency.
This gives the reader an intuitive view of the significance of each
term within the text.
The tool introduced below allows the creation of a word cloud for the set of English reviews for each spirit, as well as for the whole set of reviews composing all English reviews within the site. Credit goes to Jason Davies for the original word cloud tool which I have streamlined here for processing data output from my natural language processing (NLP) routines. Language detection was achieved via the Indico API.