Neural Network for classification with string input features

I have a classification problem where I think some kind of neural network (or maybe another machine learning approach) could be successful. For a given case, I have N string "tags" (each tag can have variable length). Given those N tags, I'd like to classify the case into one of M classes. Note that the input tags are messy and inconsistent from case to case (i.e., spaces in some cases, not others, some have underscores, different capitalization. etc). Does Matlab have features that would allow me to build a classifier with this type of input? Poking around the net, I also found the NLTK python package which may be able to help. But The rest of my program is all Matlab-based so I'd like to keep it that way if possible. Any advice or help would be much appreciated.

Thanks! Justin

[David Ding]

Hi Justin,

Off the top of my head, the following might be useful.

The new Text Analytics Toolbox might be useful:

https://uk.mathworks.com/products/text-analytics.html

As well as the LSTM networks to classify sequence data:

https://uk.mathworks.com/help/nnet/examples/classify-sequence-data-using-lstm-networks.html