3 ms·
After the recent Mechanical Turk meetup, I tried being a Mechanical Turk worker to understand the process better. One of the work units I got was classifying t
by gojomo 17y ago
After the recent Mechanical Turk meetup, I tried being a Mechanical Turk worker to understand the process better.
One of the work units I got was classifying the 'sentiments' of individual Tweets as 'positive', 'negative', 'both', 'none', or 'hard to tell'.
It seemed that many of the tweets from several work units had to do with companies in the Web2.0-finance space, and others were about people traveling by air (missing flights, waiting in airports, etc.).
I presumed I was helping to do one or more of: (1) help monitor certain direct brand/product mentions quantitatively on Twitter; (2) help do the same, indirectly, for people whose surrounding unseen tweets had linked them with specific airlines/airports; (3) help train a automated classifier to do the sentiment classifications, in general or for specific brand/experience domains.
And perhaps, the classifying fed into this Twitter offering.