Comparative Study of CNN and RNN for Natural Language Processing · HackerPapers discussions

Comparative Study of CNN and RNN for Natural Language Processing — mentioned by gtani

Sun, 02 Apr 2017 23:07:40 GMT

Things like topic recognition, entity and reference linking/disambiguation, synonym/paraphase are well researched tasks with purpose-made datasets for competitions like TREC. The most common approaches will be to choose a word embedding (one of the word2vec or Glove are good starts) and run that into a CNN or RNN variant: https://arxiv.org/abs/1702.01923 https://arxiv.org/abs/1611.09100 https://metamind.io/research/learning-when-to-skim-and-when-... https://arxiv.org/abs/1611.06792 https://arxiv.org/abs/1611.03305 probably if you search arxiv and the ACL anthology you should find good lit…

Paper and discussion history. A mention is not an endorsement.

Comparative Study of CNN and RNN for Natural Language Processing — mentioned by gtani

Mon, 20 Feb 2017 20:08:51 GMT

I think you want to understand comp linguistics viewpoint: parsers, PoS taggin, dependency analysis, syntax trees; and the machine learning perspective: embeddings in, say, 100-200 dimensional space (word2vec, glove) and topic modelling/LDA, and latent semantic analysis from the 90's. Then you can read about inputting embedding datasets into LSTM, GRU, content addressable memory/attention mechanisms etc that are being furiously introduced (you can scan the ICLR submissions and http://aclweb.org/anthology/. _____________________ The Jurafsky/Martin draft 3rd ed is a good starting point…

Paper and discussion history. A mention is not an endorsement.