Merge branch 'master' of https://github.com/yacy/yacy_search_server.git

README.md

@@ -171,8 +171,8 @@ Please clone our code and help with development!
 The code is licensed under the GPL v2.
 
 Compiling YaCy:
-- you need java 1.7 and ant
-- just compile: "ant clean all" - then you can "./startYACY.sh"
+- you need Java 1.7 or later and [Apache Ant](http://ant.apache.org/)
+- just compile: "ant clean all" - then you can "./startYACY.sh" or "./startYACY.bat"
 - create a release tarball: "ant dist"
 - create a Mac OS release: "ant distMacApp" (works only on a Mac)
 - create a debian release: "ant deb"

source/net/yacy/kelondro/data/word/WordReferenceRow.java

@@ -66,7 +66,7 @@ public final class WordReferenceRow extends AbstractReference implements WordRef
             new Column("t", Column.celltype_cardinal,  Column.encoder_b256,  2, "posintext"),
             new Column("r", Column.celltype_cardinal,  Column.encoder_b256,  1, "posinphrase"),
             new Column("o", Column.celltype_cardinal,  Column.encoder_b256,  1, "posofphrase"),
-            new Column("i", Column.celltype_cardinal,  Column.encoder_b256,  1, "worddistance"),
+            new Column("i", Column.celltype_cardinal,  Column.encoder_b256,  1, "worddistance"), // arbitrary column for avg distance of search query words
             new Column("k", Column.celltype_cardinal,  Column.encoder_b256,  1, "reserve")
         },
         Base64Order.enhancedCoder
@@ -253,7 +253,10 @@ public final class WordReferenceRow extends AbstractReference implements WordRef
     }
 
     /**
-     * First position of word in text
+     * First position of word in text.
+     * positions() is used to remember word positions for each query word of an
+     * multi word search query. As we currently don't include a separate posintext()
+     * function, we use positions to make the posintext value available.
      * @return Collection with one element
      */
     @Override

source/net/yacy/kelondro/data/word/WordReferenceVars.java

@@ -253,9 +253,7 @@ public class WordReferenceVars extends AbstractReference implements WordReferenc
                 this.hitcount,      // how often appears this word in the text
                 this.wordsintext,   // total number of words
                 this.phrasesintext, // total number of phrases
-
-                // TODO: positon 1 on empty positions may give high ranking scores for unknown pos (needs to be checked if 0 would be appropriate)
-                this.positions.isEmpty() ? -1 : this.positions.iterator().next(), // position of word in all words
+                this.positions.isEmpty() ? 0 : this.positions.iterator().next(), // position of word in all words (
                 this.posinphrase,   // position of word in its phrase
                 this.posofphrase,   // number of the phrase where word appears
                 this.lastModified,  // last-modified time of the document where word appears

source/net/yacy/kelondro/rwi/AbstractReference.java

@@ -98,17 +98,29 @@ public abstract class AbstractReference implements Reference {
         }
         return r;
     }
-    
+
+   /**
+     * max position of search query words for multi word queries
+     * @return
+     */
     @Override
     public int maxposition() {
         return max(positions());
     }
-    
+
+    /**
+     * min word position of search query words for multi word queries
+     * @return
+     */
     @Override
     public int minposition() {
         return min(positions());
     }
-    
+
+    /**
+     * The average distance (in words) between search query terms for multi word searches.
+     * @return word distance
+     */
     @Override
     public int distance() {
         if (positions().size() < 2) return 0;

source/net/yacy/kelondro/rwi/Reference.java

@@ -51,14 +51,38 @@ public interface Reference {
     @Override
     public boolean equals(Object other);
 
+    /**
+     * Joins a Reference into this one, setting the values appropriate for ranking
+     * @param joined reference
+     */
     public void join(final Reference oe);
-    
+
+    /**
+     * Positions or search query words for the referenced result url
+     * This is only valid for multi word search queries.
+     * The positions contain the first word position for every search query word
+     * which has been joined (by join() )
+     * @return list with word position
+     */
     public Collection<Integer> positions();
-    
+
+    /**
+     * max position of search query words (for multi word queries)
+     * @return
+     */
     public int maxposition();
-    
+
+    /**
+     * min word position of search query words (for multi word queries)
+     * @return
+     */
     public int minposition();
-    
+
+    /**
+     * The average distance (in words) between search query terms for multi word searches.
+     * The distance is calculated from positions()
+     * @return word distance
+     */
     public int distance();
         
 }