gistfile1.clj

(ns bktree)

(defn root [distance-fn] {:distance-fn distance-fn :children {}})
(defn new-node [term] {:term term :children {}})

(defn insert
  ([element node]
    (insert element node (:distance-fn node)))

  ([element node distance-fn]
    (if (:term node)
      (let [dist (distance-fn (:term node) element)
            children (:children node)
            child (get children dist)]

        (if child
          (assoc node :children (assoc children dist (insert element child distance-fn)))
          (assoc node :children (assoc children dist (new-node element)))))
      (assoc node :term element))))


; "distance" is some fn that takes two strings and returns an integer
; (insert "b" (insert "b" (insert "a" (root fraud/distance))))

;Sample hamming distance fn, lifted from incanter
(defn- tree-comp-each [root branch & leaves]
 (apply 
  root (map branch leaves)))

(defn- bool-to-binary [pred] (if pred 1 0))

(defn hamming-distance
  [a b]
  (if (and (integer? a) (integer? b))
    (hamming-distance (str a) (str b))
  (let [_ (assert (= (count a) (count b)))]
  (apply
   tree-comp-each 
    + 
    #(bool-to-binary (not (apply = %)))
    (map vector a b)))))