Solving a binary classification is fairly easy than to a multi-class problem. Adding more to that, when input data content is very short in nature (shorter than a tweet).
Taking a quick look on challenges here.
Are the classified outcomes statistically significant; what validation procedure makes the statistically significant test results; what accuracy can be achieved (is it reasonably higher?); Would this help in relevance search for the engine?
To answer above questions, let's visit a problem recently solved with reasonably higher accuracy.
Taking a quick look on challenges here.
Are the classified outcomes statistically significant; what validation procedure makes the statistically significant test results; what accuracy can be achieved (is it reasonably higher?); Would this help in relevance search for the engine?
To answer above questions, let's visit a problem recently solved with reasonably higher accuracy.
