A rough implementation of Normalized Google Distance in Clojure

get-ngd.clj

; Rough implementation of Normalized Google Distance algorithm
; Assumed total number of indexed pages = 42,000,000,000
(defn get-ngd
  "Returns the normalized google distance of two searchable terms. Returns nil if no results available for either query, or if there is no overlap for either query. The closer the result trends to 0, the more closely 'related' the terms are."
  [term1 term2]
  (let [m 42000000000
        fx (Integer. (:totalResults (:searchInformation (google-search term1))))
        fy (Integer. (:totalResults (:searchInformation (google-search term2))))
        fxy (Integer. (:totalResults (:searchInformation (google-search (str term1 "+" term2)))))
        ngdnumerator (- (max (math/log10 fx) (math/log10 fy)) (math/log10 fxy))
        ngddenominator (- (math/log10 m) (min (math/log10 fx) (math/log10 fy)))
        ngd (/ ngdnumerator ngddenominator)]
    (if (or (utils/NaN? ngd) (Double/isInfinite ngd)) 
      nil ngd)))