word-info.clj

; Word Information
; word: string of word
; count: # of usages
; feats: map of feature-type to feature-value
; contexts: counter of [before-word after-word] usages (for HMM)
(defrecord WordInfo [word count feats contexts])

(defn get-feats 
  "Features on a word type"
  [w]
  {:hasInitCap (boolean (re-matches #"[A-Z].*" w))    
   :hasPunc (boolean (re-matches #".*\W.*" w))
   :suffix  (let [suffix-length (min 3 (.length w))]
              (.substring #^String w (- (.length w) suffix-length)))})              

(defn new-word-info [word]
  (WordInfo. word 0 (get-feats word) (Counter. {} 0)))