A quick example of Clojure / StanfordNLP CoNLL parsing.

gistfile1.clj

(ns wujuko.nlp.fmt-read
  (:require [clojure.java.io :as io])
  (:import [java.util Properties]
           [edu.stanford.nlp.sequences
            SeqClassifierFlags
            CoNLLDocumentReaderAndWriter
            ColumnDocumentReaderAndWriter]
           [edu.stanford.nlp.ling CoreLabel]))

(defn corelabel->map
  [^CoreLabel cl]
  {:token (.word cl) :tag (.tag cl)})

(defn words
  [ms] (map :token ms))

(defn pos
  [ms] (map :tag ms))

(defn ner
  [ms] (map :ner ms))

(defn conll-seq
  ""
  [^java.io.Reader reader]
  (let [cdr (ColumnDocumentReaderAndWriter.)
        _   (.init cdr (SeqClassifierFlags.))
        ci  (.getIterator cdr reader)]
    (iterator-seq ci)))

(defn conll-map-seq
  "Returns a sequence of maps in the form:
   ( {:token \"Hi\"    :tag \"O\"}
     {:token \"There\" :tag \"O\"}
     ...
   ) "
  [^java.io.Reader reader]
  (let [cs (conll-seq reader)]
    (map (fn [ca] (map corelabel->map ca)) cs)))

(comment

  (let [filename "/Users/djunk/projects/L645/project/arktweetnlp/ark-tweet-nlp/data/twpos-data-v0.3/oct27.conll"]
    (with-open [rdr (io/reader (io/file filename))]
      (let [ms   (conll-map-seq rdr)
            samp (first (take 1 ms))]
        (words samp)
        (tags samp)
        )))

  )