I'm new in Machine learning and I'm working on a problem related to text. I know that in ML we can use features as numerical values as input to neural network, but I don't know how to use features as words. In some papers I read that we take features to be n words with some property. I really don't understand how is that possible. Please, if it is not a problem, just to tell me some good papers or textbooks or links where it is explained how to do that.
$\begingroup$
$\endgroup$
1
-
2$\begingroup$ We can make Embedding from the words, Encode them via some method Checkout TF-IDF etc Also search for Sequence Models on YouTube will Bring up Andrew NG Amazing Course $\endgroup$– AdityaCommented Mar 4, 2018 at 3:01
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
You need to make a dictionary of words. It means you have to make a dictionary which you assign each word a unique value. then you can use one-hot-encoding to represent each word uniquely. If this is what you need it will do what you want. But this has a big problem. When you think about cats and dogs, you may find similarities and differences between them. This is because you have more knowledge than the only representation of words in your brain. Consequently, you should use approaches to assign a unique number to each word, and put near concepts as neighbors. For the first part take a look at here and the second part, take a look at here.
answered Mar 4, 2018 at 5:05