Author: orbiter
Date: 2008-02-04 23:58:40 +0100 (Mon, 04 Feb 2008)
New Revision: 4445
Modified:
trunk/build.properties
trunk/htroot/Ranking_p.java
trunk/source/de/anomic/index/indexRWIEntryOrder.java
trunk/source/de/anomic/plasma/plasmaSearchRankingProcess.java
trunk/source/de/anomic/plasma/plasmaSearchRankingProfile.java
trunk/source/de/anomic/plasma/plasmaWebStructure.java
trunk/startYACY.sh
trunk/yacy.init
Log:
- some renaming of variables to support DC
- initial 120mb RAM for fresh peers
- release 0.57
Modified: trunk/build.properties
===================================================================
--- trunk/build.properties 2008-02-04 22:12:08 UTC (rev 4444)
+++ trunk/build.properties 2008-02-04 22:58:40 UTC (rev 4445)
@@ -3,7 +3,7 @@
javacTarget=1.5
# Release Configuration
-releaseVersion=0.567
+releaseVersion=0.57
stdReleaseFile=yacy_v${releaseVersion}_${DSTAMP}_${releaseNr}.tar.gz
embReleaseFile=yacy_emb_v${releaseVersion}_${DSTAMP}_${releaseNr}.tar.gz
proReleaseFile=yacy_pro_v${releaseVersion}_${DSTAMP}_${releaseNr}.tar.gz
Modified: trunk/htroot/Ranking_p.java
===================================================================
--- trunk/htroot/Ranking_p.java 2008-02-04 22:12:08 UTC (rev 4444)
+++ trunk/htroot/Ranking_p.java 2008-02-04 22:58:40 UTC (rev 4445)
@@ -44,11 +44,11 @@
private static final HashMap<String, String> rankingParameters = new
HashMap<String, String>();
static {
- rankingParameters.put(plasmaSearchRankingProfile.APPAUTHOR,
"Appearance In Author");
- rankingParameters.put(plasmaSearchRankingProfile.APPDESCR,
"Appearance In Description");
+
rankingParameters.put(plasmaSearchRankingProfile.APP_DC_CREATOR, "Appearance In
Author");
+ rankingParameters.put(plasmaSearchRankingProfile.APP_DC_TITLE,
"Appearance In Title");
rankingParameters.put(plasmaSearchRankingProfile.APPEMPH,
"Appearance In Emphasized Text");
- rankingParameters.put(plasmaSearchRankingProfile.APPREF,
"Appearance In Reference");
- rankingParameters.put(plasmaSearchRankingProfile.APPTAGS,
"Appearance In Tags");
+
rankingParameters.put(plasmaSearchRankingProfile.APP_DC_DESCRIPTION,
"Appearance In Reference/Anchor Name");
+
rankingParameters.put(plasmaSearchRankingProfile.APP_DC_SUBJECT, "Appearance In
Tags");
rankingParameters.put(plasmaSearchRankingProfile.APPURL,
"Appearance In URL");
rankingParameters.put(plasmaSearchRankingProfile.AUTHORITY,
"Authority of Domain");
rankingParameters.put(plasmaSearchRankingProfile.CATHASAPP,
"Category App, Appearance");
Modified: trunk/source/de/anomic/index/indexRWIEntryOrder.java
===================================================================
--- trunk/source/de/anomic/index/indexRWIEntryOrder.java 2008-02-04
22:12:08 UTC (rev 4444)
+++ trunk/source/de/anomic/index/indexRWIEntryOrder.java 2008-02-04
22:58:40 UTC (rev 4445)
@@ -141,18 +141,18 @@
+ ( (((t.hitcount() - min.hitcount() ) << 8) / (1 +
max.hitcount() - min.hitcount()) ) << ranking.coeff_hitcount)
+ (((int)((((t.termFrequency()- min.termFrequency() )*256.0)/ (1 +
max.termFrequency()- min.termFrequency()))))<< ranking.coeff_termfrequency)
+ ( authority(t.urlHash()) << ranking.coeff_authority)
- + (((flags.get(indexRWIEntry.flag_app_dc_identifier)) ? 255 <<
ranking.coeff_appurl : 0))
- + (((flags.get(indexRWIEntry.flag_app_dc_title)) ? 255 <<
ranking.coeff_appdescr : 0))
- + (((flags.get(indexRWIEntry.flag_app_dc_creator)) ? 255 <<
ranking.coeff_appauthor : 0))
- + (((flags.get(indexRWIEntry.flag_app_dc_subject)) ? 255 <<
ranking.coeff_apptags : 0))
- + (((flags.get(indexRWIEntry.flag_app_dc_description)) ? 255 <<
ranking.coeff_appref : 0))
- + (((flags.get(indexRWIEntry.flag_app_emphasized)) ? 255 <<
ranking.coeff_appemph : 0))
- + (((flags.get(plasmaCondenser.flag_cat_indexof)) ? 255 <<
ranking.coeff_catindexof : 0))
- + (((flags.get(plasmaCondenser.flag_cat_hasimage)) ? 255 <<
ranking.coeff_cathasimage : 0))
- + (((flags.get(plasmaCondenser.flag_cat_hasaudio)) ? 255 <<
ranking.coeff_cathasaudio : 0))
- + (((flags.get(plasmaCondenser.flag_cat_hasvideo)) ? 255 <<
ranking.coeff_cathasvideo : 0))
- + (((flags.get(plasmaCondenser.flag_cat_hasapp)) ? 255 <<
ranking.coeff_cathasapp : 0))
- + (((yacyURL.probablyRootURL(t.urlHash())) ? 15 <<
ranking.coeff_urllength : 0));
+ + (((flags.get(indexRWIEntry.flag_app_dc_identifier)) ? 255 <<
ranking.coeff_appurl : 0))
+ + (((flags.get(indexRWIEntry.flag_app_dc_title)) ? 255 <<
ranking.coeff_app_dc_title : 0))
+ + (((flags.get(indexRWIEntry.flag_app_dc_creator)) ? 255 <<
ranking.coeff_app_dc_creator : 0))
+ + (((flags.get(indexRWIEntry.flag_app_dc_subject)) ? 255 <<
ranking.coeff_app_dc_subject : 0))
+ + (((flags.get(indexRWIEntry.flag_app_dc_description)) ? 255 <<
ranking.coeff_app_dc_description : 0))
+ + (((flags.get(indexRWIEntry.flag_app_emphasized)) ? 255 <<
ranking.coeff_appemph : 0))
+ + (((flags.get(plasmaCondenser.flag_cat_indexof)) ? 255 <<
ranking.coeff_catindexof : 0))
+ + (((flags.get(plasmaCondenser.flag_cat_hasimage)) ? 255 <<
ranking.coeff_cathasimage : 0))
+ + (((flags.get(plasmaCondenser.flag_cat_hasaudio)) ? 255 <<
ranking.coeff_cathasaudio : 0))
+ + (((flags.get(plasmaCondenser.flag_cat_hasvideo)) ? 255 <<
ranking.coeff_cathasvideo : 0))
+ + (((flags.get(plasmaCondenser.flag_cat_hasapp)) ? 255 <<
ranking.coeff_cathasapp : 0))
+ + (((yacyURL.probablyRootURL(t.urlHash())) ? 15 <<
ranking.coeff_urllength : 0));
//if (searchWords != null) r += (yacyURL.probablyWordURL(t.urlHash(),
searchWords) != null) ? 256 << ranking.coeff_appurl : 0;
return Long.MAX_VALUE - r; // returns a reversed number: the lower the
number the better the ranking. This is used for simple sorting with a TreeMap
Modified: trunk/source/de/anomic/plasma/plasmaSearchRankingProcess.java
===================================================================
--- trunk/source/de/anomic/plasma/plasmaSearchRankingProcess.java
2008-02-04 22:12:08 UTC (rev 4444)
+++ trunk/source/de/anomic/plasma/plasmaSearchRankingProcess.java
2008-02-04 22:58:40 UTC (rev 4445)
@@ -521,7 +521,7 @@
while (shi.hasNext()) {
queryhash = shi.next();
if (urlcomph.contains(queryhash)) r += 256 <<
query.ranking.coeff_appurl;
- if (descrcomph.contains(queryhash)) r += 256 <<
query.ranking.coeff_appdescr;
+ if (descrcomph.contains(queryhash)) r += 256 <<
query.ranking.coeff_app_dc_title;
}
return r;
Modified: trunk/source/de/anomic/plasma/plasmaSearchRankingProfile.java
===================================================================
--- trunk/source/de/anomic/plasma/plasmaSearchRankingProfile.java
2008-02-04 22:12:08 UTC (rev 4444)
+++ trunk/source/de/anomic/plasma/plasmaSearchRankingProfile.java
2008-02-04 22:58:40 UTC (rev 4445)
@@ -48,34 +48,34 @@
public class plasmaSearchRankingProfile {
// pre-sort attributes
- public static final String DOMLENGTH = "domlength";
- public static final String YBR = "ybr";
- public static final String DATE = "date";
- public static final String WORDSINTITLE = "wordsintitle";
- public static final String WORDSINTEXT = "wordsintext";
- public static final String PHRASESINTEXT = "phrasesintext";
- public static final String LLOCAL = "llocal";
- public static final String LOTHER = "lother";
- public static final String URLLENGTH = "urllength";
- public static final String URLCOMPS = "urlcomps";
- public static final String HITCOUNT = "hitcount";
- public static final String POSINTEXT = "posintext";
- public static final String POSOFPHRASE = "posofphrase";
- public static final String POSINPHRASE = "posinphrase";
- public static final String AUTHORITY = "authority";
- public static final String WORDDISTANCE = "worddistance";
- public static final String APPURL = "appurl";
- public static final String APPDESCR = "appdescr";
- public static final String APPAUTHOR = "appauthor";
- public static final String APPTAGS = "apptags";
- public static final String APPREF = "appref";
- public static final String APPEMPH = "appemph";
- public static final String CATINDEXOF = "catindexof";
- public static final String CATHASIMAGE = "cathasimage";
- public static final String CATHASAUDIO = "cathasaudio";
- public static final String CATHASVIDEO = "cathasvideo";
- public static final String CATHASAPP = "cathasapp";
- public static final String TERMFREQUENCY = "tf";
+ public static final String DOMLENGTH = "domlength";
+ public static final String YBR = "ybr";
+ public static final String DATE = "date";
+ public static final String WORDSINTITLE = "wordsintitle";
+ public static final String WORDSINTEXT = "wordsintext";
+ public static final String PHRASESINTEXT = "phrasesintext";
+ public static final String LLOCAL = "llocal";
+ public static final String LOTHER = "lother";
+ public static final String URLLENGTH = "urllength";
+ public static final String URLCOMPS = "urlcomps";
+ public static final String HITCOUNT = "hitcount";
+ public static final String POSINTEXT = "posintext";
+ public static final String POSOFPHRASE = "posofphrase";
+ public static final String POSINPHRASE = "posinphrase";
+ public static final String AUTHORITY = "authority";
+ public static final String WORDDISTANCE = "worddistance";
+ public static final String APPURL = "appurl";
+ public static final String APP_DC_TITLE = "appdescr"; // title of
page
+ public static final String APP_DC_CREATOR = "appauthor"; // the
author field
+ public static final String APP_DC_SUBJECT = "apptags"; // tags
+ public static final String APP_DC_DESCRIPTION = "appref"; //
references to the source (content of <a> tag)
+ public static final String APPEMPH = "appemph";
+ public static final String CATINDEXOF = "catindexof";
+ public static final String CATHASIMAGE = "cathasimage";
+ public static final String CATHASAUDIO = "cathasaudio";
+ public static final String CATHASVIDEO = "cathasvideo";
+ public static final String CATHASAPP = "cathasapp";
+ public static final String TERMFREQUENCY = "tf";
// post-sort predicates
public static final String URLCOMPINTOPLIST = "urlcompintoplist";
@@ -90,7 +90,7 @@
coeff_domlength, coeff_ybr, coeff_date, coeff_wordsintitle,
coeff_wordsintext, coeff_phrasesintext,
coeff_llocal, coeff_lother, coeff_urllength, coeff_urlcomps,
coeff_hitcount,
coeff_posintext, coeff_posofphrase, coeff_posinphrase,
coeff_authority, coeff_worddistance,
- coeff_appurl, coeff_appdescr, coeff_appauthor, coeff_apptags,
coeff_appref, coeff_appemph,
+ coeff_appurl, coeff_app_dc_title, coeff_app_dc_creator,
coeff_app_dc_subject, coeff_app_dc_description, coeff_appemph,
coeff_catindexof, coeff_cathasimage, coeff_cathasaudio,
coeff_cathasvideo, coeff_cathasapp,
coeff_urlcompintoplist, coeff_descrcompintoplist, coeff_prefer,
coeff_termfrequency;
@@ -114,10 +114,10 @@
coeff_authority = 11;
coeff_worddistance = 12;
coeff_appurl = 13;
- coeff_appdescr = 10;
- coeff_appauthor = 11;
- coeff_apptags = 8;
- coeff_appref = 7;
+ coeff_app_dc_title = 13;
+ coeff_app_dc_creator = 12;
+ coeff_app_dc_subject = 9;
+ coeff_app_dc_description = 8;
coeff_appemph = 10;
coeff_catindexof = (mediatype ==
plasmaSearchQuery.CONTENTDOM_TEXT) ? 0 : 15;
coeff_cathasimage = (mediatype ==
plasmaSearchQuery.CONTENTDOM_IMAGE) ? 15 : 0;
@@ -127,7 +127,7 @@
coeff_termfrequency = 14;
coeff_urlcompintoplist = 3;
coeff_descrcompintoplist = 2;
- coeff_prefer = 12;
+ coeff_prefer = 14;
}
public plasmaSearchRankingProfile(String prefix, String profile) {
@@ -165,10 +165,10 @@
coeff_authority = parseMap(coeff, AUTHORITY,
coeff_authority);
coeff_worddistance = parseMap(coeff, WORDDISTANCE,
coeff_worddistance);
coeff_appurl = parseMap(coeff, APPURL, coeff_appurl);
- coeff_appdescr = parseMap(coeff, APPDESCR,
coeff_appdescr);
- coeff_appauthor = parseMap(coeff, APPAUTHOR,
coeff_appauthor);
- coeff_apptags = parseMap(coeff, APPTAGS, coeff_apptags);
- coeff_appref = parseMap(coeff, APPREF, coeff_appref);
+ coeff_app_dc_title = parseMap(coeff, APP_DC_TITLE,
coeff_app_dc_title);
+ coeff_app_dc_creator = parseMap(coeff, APP_DC_CREATOR,
coeff_app_dc_creator);
+ coeff_app_dc_subject = parseMap(coeff, APP_DC_SUBJECT,
coeff_app_dc_subject);
+ coeff_app_dc_description = parseMap(coeff, APP_DC_DESCRIPTION,
coeff_app_dc_description);
coeff_appemph = parseMap(coeff, APPEMPH, coeff_appemph);
coeff_catindexof = parseMap(coeff, CATINDEXOF,
coeff_catindexof);
coeff_cathasimage = parseMap(coeff, CATHASIMAGE,
coeff_cathasimage);
@@ -221,10 +221,10 @@
ext.put(prefix + AUTHORITY, Integer.toString(coeff_authority));
ext.put(prefix + WORDDISTANCE, Integer.toString(coeff_worddistance));
ext.put(prefix + APPURL, Integer.toString(coeff_appurl));
- ext.put(prefix + APPDESCR, Integer.toString(coeff_appdescr));
- ext.put(prefix + APPAUTHOR, Integer.toString(coeff_appauthor));
- ext.put(prefix + APPTAGS, Integer.toString(coeff_apptags));
- ext.put(prefix + APPREF, Integer.toString(coeff_appref));
+ ext.put(prefix + APP_DC_TITLE, Integer.toString(coeff_app_dc_title));
+ ext.put(prefix + APP_DC_CREATOR,
Integer.toString(coeff_app_dc_creator));
+ ext.put(prefix + APP_DC_SUBJECT,
Integer.toString(coeff_app_dc_subject));
+ ext.put(prefix + APP_DC_DESCRIPTION,
Integer.toString(coeff_app_dc_description));
ext.put(prefix + APPEMPH, Integer.toString(coeff_appemph));
ext.put(prefix + CATINDEXOF, Integer.toString(coeff_catindexof));
ext.put(prefix + CATHASIMAGE, Integer.toString(coeff_cathasimage));
Modified: trunk/source/de/anomic/plasma/plasmaWebStructure.java
===================================================================
--- trunk/source/de/anomic/plasma/plasmaWebStructure.java 2008-02-04
22:12:08 UTC (rev 4444)
+++ trunk/source/de/anomic/plasma/plasmaWebStructure.java 2008-02-04
22:58:40 UTC (rev 4445)
@@ -67,7 +67,7 @@
this.structureFile = structureFile;
// load web structure
- Map<String, String> loadedStructure =
serverFileUtils.loadHashMap(this.structureFile);
+ Map<String, String> loadedStructure = (this.structureFile.exists()) ?
serverFileUtils.loadHashMap(this.structureFile) : new TreeMap<String, String>();
if (loadedStructure != null) this.structure.putAll(loadedStructure);
// delete outdated entries in case the structure is too big
Modified: trunk/startYACY.sh
===================================================================
--- trunk/startYACY.sh 2008-02-04 22:12:08 UTC (rev 4444)
+++ trunk/startYACY.sh 2008-02-04 22:58:40 UTC (rev 4445)
@@ -96,6 +96,8 @@
# i="${i#javastart_*=}";
# JAVA_ARGS="-$i $JAVA_ARGS";
# done
+else
+ JAVA_ARGS="-Xmx120m -Xms120m $JAVA_ARGS";
fi
#echo "JAVA_ARGS: $JAVA_ARGS"
#echo "JAVA: $JAVA"
Modified: trunk/yacy.init
===================================================================
--- trunk/yacy.init 2008-02-04 22:12:08 UTC (rev 4444)
+++ trunk/yacy.init 2008-02-04 22:58:40 UTC (rev 4445)
@@ -640,10 +640,8 @@
# -Xmx<size> and -Xms<size> maximum/init Java heap size
# both values should be equal,
# othervise the YaCy-internal memory supervision does not work
-javastart_Xmx=Xmx96m
-javastart_Xms=Xms96m
-javastart_Xmx__pro=Xmx120m
-javastart_Xms__pro=Xms120m
+javastart_Xmx=Xmx120m
+javastart_Xms=Xms120m
# priority of the yacy-process
# is valid in unix/shell and windows environments but
_______________________________________________
YaCy-svn mailing list
YaCy-svn@lists.berlios.de
https://lists.berlios.de/mailman/listinfo/yacy-svn
