Rogan Creswick wrote:
Haskell has very limited support for high-level Natural Language Processing (tokenization, sentence splitting, Named-entity recognition, etc...).
Since the role of a general purpose language is relatively new for Haskell, there are many areas where Haskell is still an emerging language. So it is interesting that you say that about NLP, where Haskell is not only quite mature, but arguably the leading language today. For example, the EU's huge Molto project http://molto-project.eu/, which aims to provide automated real-time high-quality translation of a wide class of documents between all of the EU languages, is based on Haskell's GF http://grammaticalframework.org/. Coincidentally, just yesterday a company that markets one of the top semantic NLP products contacted me. They have decided to dump their entire Java code base, using older technologies such as the ones you mentioned. One of their leading candidates for a replacement language is Haskell. They told me, "You see Haskell everywhere in NLP these days." See http://www.haskell.org/haskellwiki/Applications_and_libraries/Linguistics for links about NLP work in Haskell. That is a huge wiki page and hard to maintain, so some of the links are out of date, but you get the idea. Regards, Yitz