Thanks for coming, but this post is not yet ready. …Being written as we speak, come back again!
NLI and machine translation.
Why is NLI important ? It’s a key problem in natural language understanding. Many other NLU problems can be reduced to NLI: such as summarization (given a piece of text and a suggested summary, does the former entail the latter), information extraction (does the text entail the extracted fact), question answering (does the data source entail a given question and answer pair) as well as machine machine translation (does a phrase in language A entail its given translation in language B and vice versa). Why is it mentioned in the same category as machine translation? Both can be cast as an alignment problem.
Question Answering
Reading comprehension, memory and attention.
Dialogue and chatbots!
So why hs DL been more successful in ASR and vision than NLP/NLU? That’s a topic for another post!