arquLV

3 ms·

1. 1.▲
   by arquLV 9y ago

   The samples were selected randomly at the beginning. Tags were thought up on the go, trying to generalize into broader categories. After the initial tagging, we counted the number of samples for each category and grouped similar underrepre
2. 2.▲
   by arquLV 9y ago

   Yes, we're analyzing the titles. They're vectorized and fed into a LSTM net which was trained on a manually tagged training set. The set is not very big yet but yields good enough results for the initial proof of concept.
3. 3.▲
   by arquLV 9y ago

   Hey, thanks! It's a bit of deep learning magic combined with a few days of manual labor tagging the training set :) The classifier itself is an LSTM and runs on TensorFlow. As said, this is a proof of concept so we'll try to impr