Increasing NLP Parsing Efficiency with Chunking
Mark Dáibhidh Anderson, David Vilares · 2018
We introduce a “Chunk-and-Pass” parsing technique influenced by a psycholinguistic model, where linguistic information is processed not word-by-word but rather in larger chunks of words. We present preliminary results that show that it is feasible to compress linguistic data into chunks without significantly diminishing parsing performance and potentially increasing the speed.