Multiclassification with DLS

Hi I am trying to set up a sentiment classifier that needs to classify text with different sentiments. The data can take more than one sentiment at a time.

I am wondering how to set up my train.csv file. I am currently passing the text as one column of semicolon separated numbers, created with keras tokenizer. Not sure how to pass the sentiment classes though. I tried multiple columns, one column with comma separated and one-hot-encoding.

Thanks in advance.

Kind regards,

Hi Theodore,

You can try with single column for sentiments. You can choose array data type. Array data type is encoded as sequence of numbers separated using semicolon. For example, if you have 3 sentiment classes, you can encode it as one hot vector separated by semicolons (1;0;1)


1 Like

Thanks Rajendra,

Will give it a try!

Kind regards,