1
$\begingroup$

So I have this situation, I have filtered a bunch of single independent sentences that I filtered because they contain the word X (in my case, X = "budget"). if the meaning of the sentence is "budget increases/goes up/etc" my result is 1 and, if not, 0. all sentences hold such meaning (the word budget is not used in different ways)

this is not a traditional "sentiment" problem and I do not know how to approach this. I could run into a variety of cases ("budget did not go down" // "A budget cut was in order" // "A budget cut was not in order")

if required, I can categorize by hand up to 400/500 sentences and say with certainty the result (0, 1). could you point me to any model that would solve this? would the amount of hand-labeled samples be enough?

$\endgroup$
0
$\begingroup$

This seems like a straight forward classification problem: "Increased", "Decreased" (I assume "Neither Increased nor Decreased" can be a third category as well). Something simple like a Support Vector Machine (SVM) could work.

That being said, only 400-500 example cases aren't a lot so you might run into overfitting especially if you try to run whatever model you land on later down the road.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.