From b632667ce57af89691407bb8668e1512775278ae Mon Sep 17 00:00:00 2001
From: Calvin <calvin@EESI>
Date: Fri, 15 Mar 2013 15:26:20 -0400
Subject: nbc added

---
 src/nbc/score.sml | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)
 create mode 100644 src/nbc/score.sml

(limited to 'src/nbc/score.sml')

diff --git a/src/nbc/score.sml b/src/nbc/score.sml
new file mode 100644
index 0000000..90d569a
--- /dev/null
+++ b/src/nbc/score.sml
@@ -0,0 +1,30 @@
+signature SCORE = sig
+	val score: int * real * (string -> int option) * real * string -> real
+end
+
+structure Score :> SCORE = struct
+	fun |> (x, f) = f x
+	infix |>
+
+	fun addCount (hitsum, fcount, gcount, totalWords) =
+		Kahan.add (hitsum, Real.fromInt fcount * Math.ln (Real.fromInt gcount / totalWords))
+	fun addNmer (totalWords, getGenomeCount) (nmer, ref fcount, (misses, anyhits, hitsum)) =
+		case getGenomeCount nmer of
+			NONE => (misses + 1, anyhits, hitsum)
+			| SOME gcount => (
+				misses, true
+				, addCount (hitsum, fcount, gcount, totalWords)
+			)
+	fun score (order, missConstant, getGenomeCount, totalWords, fragment) =
+		let
+			val add = addNmer (totalWords, getGenomeCount)
+			val seed = (0, false, Kahan.zero)
+			val (misses, anyhits, hitsum) =
+				Nmer.count (order, fragment) |> HashTable.foldi add seed
+		in
+			if anyhits then
+				Kahan.add (hitsum, Math.ln missConstant * Real.fromInt misses)
+				|> Kahan.sum
+			else Real.negInf
+		end
+end
-- 
cgit v1.2.3