Record scores of distillability and long article in UMA

components/dom_distiller/content/renderer/distillability_agent.cc

Index: components/dom_distiller/content/renderer/distillability_agent.cc
diff --git a/components/dom_distiller/content/renderer/distillability_agent.cc b/components/dom_distiller/content/renderer/distillability_agent.cc
index f0ae4b511a4b0d86e8ea6bc99f5709f17e4d8d03..f2533e918ca6dcaf54417929263b10018f8c6bed 100644
--- a/components/dom_distiller/content/renderer/distillability_agent.cc
+++ b/components/dom_distiller/content/renderer/distillability_agent.cc
@@ -93,10 +93,33 @@ bool IsDistillablePageAdaboost(WebDocument& doc,
     features.mozScoreAllSqrt,
     features.mozScoreAllLinear
   );
-  bool distillable = detector->Classify(derived);
-  bool long_article = long_page->Classify(derived);
+  double score = detector->Score(derived) - detector->GetThreshold();
+  double long_score = long_page->Score(derived) - long_page->GetThreshold();
+  bool distillable = score > 0;
+  bool long_article = long_score > 0;
   bool blacklisted = IsBlacklisted(parsed_url);
 
+  if (!features.isMobileFriendly) {
+    int score_int = std::round(score * 100);
+    if (score > 0) {
+      UMA_HISTOGRAM_COUNTS_1000("DomDistiller.DistillabilityScoreNMF.Positive",
+          score_int);
+    } else {
+      UMA_HISTOGRAM_COUNTS_1000("DomDistiller.DistillabilityScoreNMF.Negative",
+          -score_int);
+    }
+    if (distillable) {

[rkaplow, 2016/06/07 20:51:12]

this is confusing me, since the above section says Distillability metrics.

[wychen, 2016/06/07 21:27:33]

The long-page score is only relevant if the first model (|detector|) says yes,
since it is trained this way.
Therefore, we only collect the distribution in this case.

+      int long_score_int = std::round(long_score * 100);
+      if (long_score > 0) {
+        UMA_HISTOGRAM_COUNTS_1000("DomDistiller.LongArticleScoreNMF.Positive",
+            long_score_int);
+      } else {
+        UMA_HISTOGRAM_COUNTS_1000("DomDistiller.LongArticleScoreNMF.Negative",
+            -long_score_int);
+      }
+    }
+  }
+
   int bucket = static_cast<unsigned>(features.isMobileFriendly) |
       (static_cast<unsigned>(distillable) << 1);
   if (is_last) {