From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Thu, 15 Nov 2018 19:34:20 +0000 (+0000)
Subject: [Fix] Switch from chi-square to naive for large Fisher value
X-Git-Tag: 1.8.2~23
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=df529a8d06aa25285362175c75c5275b9566b0b1;p=rspamd.git

[Fix] Switch from chi-square to naive for large Fisher value
---

diff --git a/src/libstat/classifiers/bayes.c b/src/libstat/classifiers/bayes.c
index 1a28a42cb..2e710522b 100644
--- a/src/libstat/classifiers/bayes.c
+++ b/src/libstat/classifiers/bayes.c
@@ -344,8 +344,24 @@ bayes_classify (struct rspamd_classifier * ctx,
 		return TRUE;
 	}
 
-	h = 1 - inv_chi_square (task, cl.spam_prob, cl.processed_tokens);
-	s = 1 - inv_chi_square (task, cl.ham_prob, cl.processed_tokens);
+	if (cl.spam_prob < -300 && cl.ham_prob < -300) {
+		/* Fisher value is low enough to apply inv_chi_square */
+		h = 1 - inv_chi_square (task, cl.spam_prob, cl.processed_tokens);
+		s = 1 - inv_chi_square (task, cl.ham_prob, cl.processed_tokens);
+	}
+	else {
+		/* Use naive method */
+		if (cl.spam_prob > cl.ham_prob) {
+			s = (1.0 - exp(cl.spam_prob / cl.ham_prob)) /
+					(1.0 + exp(cl.spam_prob / cl.ham_prob));
+			h = 1.0 - s;
+		}
+		else {
+			h = (1.0 - exp(cl.ham_prob / cl.spam_prob)) /
+				(1.0 + exp(cl.ham_prob / cl.spam_prob));
+			s = 1.0 - h;
+		}
+	}
 
 	if (isfinite (s) && isfinite (h)) {
 		final_prob = (s + 1.0 - h) / 2.;