From 365a3fff8e48a73f6c139f8a3c00e8a8449011e1 Mon Sep 17 00:00:00 2001
From: orbiter <orbiter@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Tue, 7 Feb 2006 23:16:46 +0000
Subject: [PATCH] fixings for ranking attributes

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@1569 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 .../plasma/plasmaSearchRankingProfile.java       | 16 ++++++++--------
 source/de/anomic/plasma/plasmaURL.java           |  2 +-
 .../de/anomic/plasma/plasmaWordIndexEntry.java   |  2 +-
 3 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/source/de/anomic/plasma/plasmaSearchRankingProfile.java b/source/de/anomic/plasma/plasmaSearchRankingProfile.java
index 97809a0af..8cc4a98b5 100644
--- a/source/de/anomic/plasma/plasmaSearchRankingProfile.java
+++ b/source/de/anomic/plasma/plasmaSearchRankingProfile.java
@@ -185,10 +185,10 @@ public class plasmaSearchRankingProfile {
 
         // apply 'common-sense' heuristic using references
         for (int j = 0; j < urlcomps.length; j++) {
-            if (topwords.contains(urlcomps[j])) ranking += 1 << ((Integer) coeff.get(URLCOMPINTOPLIST)).intValue();
+            if (topwords.contains(urlcomps[j])) ranking += 256 << ((Integer) coeff.get(URLCOMPINTOPLIST)).intValue();
         }
         for (int j = 0; j < descrcomps.length; j++) {
-            if (topwords.contains(descrcomps[j])) ranking += 1 << ((Integer) coeff.get(DESCRCOMPINTOPLIST)).intValue();
+            if (topwords.contains(descrcomps[j])) ranking += 256 << ((Integer) coeff.get(DESCRCOMPINTOPLIST)).intValue();
         }
 
         // apply query-in-result matching
@@ -198,17 +198,17 @@ public class plasmaSearchRankingProfile {
         String queryhash;
         while (shi.hasNext()) {
             queryhash = (String) shi.next();
-            if (urlcomph.contains(queryhash)) ranking += 1 << ((Integer) coeff.get(QUERYINURL)).intValue();
-            if (descrcomph.contains(queryhash)) ranking += 1 << ((Integer) coeff.get(QUERYINDESCR)).intValue();
+            if (urlcomph.contains(queryhash)) ranking += 256 << ((Integer) coeff.get(QUERYINURL)).intValue();
+            if (descrcomph.contains(queryhash)) ranking += 256 << ((Integer) coeff.get(QUERYINDESCR)).intValue();
         }
 
         // prefer short urls
-        ranking += (255 - page.url().toString().length()) << ((Integer) coeff.get(URLLENGTH)).intValue();
-        ranking += (24 - urlcomps.length) << ((Integer) coeff.get(URLCOMPS)).intValue();
+        ranking += (256 - page.url().toString().length()) << ((Integer) coeff.get(URLLENGTH)).intValue();
+        ranking += (32 - urlcomps.length) << ((Integer) coeff.get(URLCOMPS)).intValue();
 
         // prefer long descriptions
-        ranking += (40 - Math.abs(40 - Math.min(40, page.descr().length()))) << ((Integer) coeff.get(DESCRLENGTH)).intValue();
-        ranking += (8 - Math.abs(8 - Math.min(8, descrcomps.length))) << ((Integer) coeff.get(DESCRCOMPS)).intValue();
+        ranking += (255 * page.descr().length() / 80) << ((Integer) coeff.get(DESCRLENGTH)).intValue();
+        ranking += (255 * (12 - Math.abs(12 - Math.min(12, descrcomps.length))) / 12) << ((Integer) coeff.get(DESCRCOMPS)).intValue();
 
         return ranking;
     }
diff --git a/source/de/anomic/plasma/plasmaURL.java b/source/de/anomic/plasma/plasmaURL.java
index b26424eb9..650d47d9d 100644
--- a/source/de/anomic/plasma/plasmaURL.java
+++ b/source/de/anomic/plasma/plasmaURL.java
@@ -538,7 +538,7 @@ public class plasmaURL {
     public static final int domLengthEstimation(String urlHash) {
         // generates an estimation of the original domain length
         int flagbyte = kelondroBase64Order.enhancedCoder.decodeByte(urlHash.charAt(11));
-        int domLengthKey = flagbyte & 4;
+        int domLengthKey = flagbyte & 3;
         switch (domLengthKey) {
             case 0: return 4;
             case 1: return 10;
diff --git a/source/de/anomic/plasma/plasmaWordIndexEntry.java b/source/de/anomic/plasma/plasmaWordIndexEntry.java
index fa215a6bc..5f5e127c2 100644
--- a/source/de/anomic/plasma/plasmaWordIndexEntry.java
+++ b/source/de/anomic/plasma/plasmaWordIndexEntry.java
@@ -410,7 +410,7 @@ public final class plasmaWordIndexEntry implements Cloneable {
     public boolean isLocal() { return localflag == LT_LOCAL; }
     
     public int domlengthNormalized() {
-        return 255 * plasmaURL.domLengthEstimation(this.urlHash) / 20;
+        return 255 * plasmaURL.domLengthEstimation(this.urlHash) / 30;
     }
 
     public static void main(String[] args) {