[Feature] Try to improve normalization function for bayes

diff --git a/src/libutil/util.c b/src/libutil/util.c
index 5380d8c9443125538787471cb867ef415b8b8ce2..d1332d50478c244a9a7c748db7f32f7fe8e7e68e 100644 (file)
--- a/src/libutil/util.c
+++ b/src/libutil/util.c
@@ -2549,17 +2549,16 @@ rspamd_shmem_xmap (const char *fname, guint mode,
  * C = -7
  * D = 3
  * y = 32(x - 0.5)^4 - 6(x - 0.5)^3 - 7(x - 0.5)^2 + 3(x - 0.5)
+ *
+ * New approach:
+ * y = ((x - bias)*2)^8
  */
 gdouble
 rspamd_normalize_probability (gdouble x, gdouble bias)
 {
-       const gdouble a = 32, b = -6, c = -7, d = 3;
-       gdouble xx, x2, x3, x4;
+       gdouble xx;
 
-       xx = x - bias;
-       x2 = xx * xx;
-       x3 = x2 * xx;
-       x4 = x3 * xx;
+       xx = (x - bias) * 2.0;
 
-       return a*x4 + b*x3 + c*x2 + d*xx;
+       return pow (xx, 8);
 }

diff --git a/src/lua/lua_util.c b/src/lua/lua_util.c
index cb24cb983b22b96cce7ace38f5a0a6f0b277fd80..8aec3a88d80100756968517a6700bc013f178a14 100644 (file)
--- a/src/lua/lua_util.c
+++ b/src/lua/lua_util.c
@@ -352,6 +352,16 @@ LUA_FUNCTION_DEF (util, zstd_compress);
  */
 LUA_FUNCTION_DEF (util, zstd_decompress);
 
+/***
+ * @function util.normalize_prob(prob, [bias = 0.5])
+ * Normalize probabilities using polynom
+ *
+ * @param {number} prob probability param
+ * @param {number} bias number to subtract for making the final sollution
+ * @return {number} normalized number
+ */
+LUA_FUNCTION_DEF (util, normalize_prob);
+
 /***
  * @function util.pack(fmt, ...)
  *
@@ -463,6 +473,7 @@ static const struct luaL_reg utillib_f[] = {
        LUA_INTERFACE_DEF (util, random_hex),
        LUA_INTERFACE_DEF (util, zstd_compress),
        LUA_INTERFACE_DEF (util, zstd_decompress),
+       LUA_INTERFACE_DEF (util, normalize_prob),
        LUA_INTERFACE_DEF (util, pack),
        LUA_INTERFACE_DEF (util, unpack),
        LUA_INTERFACE_DEF (util, packsize),
@@ -1724,6 +1735,23 @@ lua_util_zstd_decompress (lua_State *L)
        return 2;
 }
 
+static gint
+lua_util_normalize_prob (lua_State *L)
+{
+       gdouble x, bias = 0.5;
+
+       x = lua_tonumber (L, 1);
+
+       if (lua_type (L, 2) == LUA_TNUMBER) {
+               bias = lua_tonumber (L, 2);
+       }
+
+       lua_pushnumber (L, rspamd_normalize_probability (x, bias));
+
+       return 1;
+}
+
+
 /* Backport from Lua 5.3 */
 
 /******************************************************************************