Graphics Reference
In-Depth Information
outside of the hard-core fan base, where newbie fans are more likely to be
drawn in, and the effect of that excitement seems to be viral.
Community Sentiment
Mapping color to topics has shown you what communities are talking about
and revealed patterns among topics. But what if you want to know what
they are
saying
about those topics? Are the fans in this data set raving
about the players, trashing them, or something in between? Determining
whether a text is likely positive, negative, or neutral is known as
sentiment
classification
.
One way to perform sentiment classification is to use a computer process
whereby an algorithm that has been previously “trained” by a person
estimates the sentiment of a new text based on similarity to text on which it
was trained. This approach is often ideal in that the algorithm is capable of
learningbyexamplewithoutexhaustivesemanticinstructionfromahuman.
Butforresultsofreasonablequality,itrequiresthatthealgorithmbetrained
on similar data. For this data set, you want a classifier that has been trained
specifically for social media.
For simplicity, this example uses a convenient online web service provided
by Viral Heat to classify fan comments directly from Excel. If you don't
already have a Viral Heat account, you will need to sign up for one at
https://app.viralheat.com/developer/
and request a developer API key
online, or use an alternate service.
Reopen the Raptors fan spreadsheet you enhanced with player mentions,
and start by adding a new row to the top of the
Vertices
worksheet,
followed by four empty columns to the right of the Tweet column. Name
the new columns “Mood,” “Mood Probability,” “Sentiment,” and “Topic
Symbols.” To the far right, outside of the graph data area, add four new
columns where the sentiment will be computed, and a fifth that will be
reserved for visually expressing topics. Label these new columns
“Classification,” “Mood Calc,” “Mood Probability Calc,” “Sentiment Calc,”
and “Topic Symbols Calc.”
In the Classification column, add the following formula to the vertex rows,
replacing
your_key_here
with your own API key:
=WEBSERVICE("https://app.viralheat.com/social/api/
sentiment?text=
