From 1d4e206b2bc34a70db8ea3460abc5907766f9e5c Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Mon, 18 Jun 2012 18:10:40 +0200
Subject: [PATCH] bugfix in vocabulary generation

---
 htroot/Vocabulary_p.java                         | 8 ++++----
 source/net/yacy/cora/lod/vocabulary/Tagging.java | 2 +-
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/htroot/Vocabulary_p.java b/htroot/Vocabulary_p.java
index 6021e9c7c..68064fb1b 100644
--- a/htroot/Vocabulary_p.java
+++ b/htroot/Vocabulary_p.java
@@ -96,14 +96,14 @@ public class Vocabulary_p {
                                     URIMetadataRow m = segment.urlMetadata().load(u.hash());
                                     if (m != null) t = m.dc_creator();
                                 }
-                                t = t.replaceAll("\"", " ").replaceAll("'", " ").replaceAll(",", " ").replaceAll("  ", " ").trim();
+                                t = t.replaceAll("_", " ").replaceAll("\"", " ").replaceAll("'", " ").replaceAll(",", " ").replaceAll("  ", " ").trim();
                                 if (t.length() == 0) continue;
                                 if (discoverFromTitleSplitted) {
                                     String[] ts = t.split(" ");
                                     for (String s: ts) {
                                         if (s.length() == 0) continue;
                                         if (s.endsWith(".jpg") || s.endsWith(".gif")) continue;
-                                        table.put(s, new Tagging.SOTuple("", u0));
+                                        table.put(s, new Tagging.SOTuple(Tagging.normalizeTerm(s), u0));
                                     }
                                 } else if (discoverFromAuthor) {
                                     String[] ts = t.split(";"); // author names are often separated by ';'
@@ -111,10 +111,10 @@ public class Vocabulary_p {
                                         if (s.length() == 0) continue;
                                         int p = s.indexOf(','); // check if there is a reversed method to mention the name
                                         if (p >= 0) s = s.substring(p + 1).trim() + " " + s.substring(0, p).trim();
-                                        table.put(s, new Tagging.SOTuple("", u0));
+                                        table.put(s, new Tagging.SOTuple(Tagging.normalizeTerm(s), u0));
                                     }
                                 } else {
-                                    table.put(t, new Tagging.SOTuple("", u0));
+                                    table.put(t, new Tagging.SOTuple(Tagging.normalizeTerm(t), u0));
                                 }
                             }
                         }
diff --git a/source/net/yacy/cora/lod/vocabulary/Tagging.java b/source/net/yacy/cora/lod/vocabulary/Tagging.java
index 3239deb48..6a931eb88 100644
--- a/source/net/yacy/cora/lod/vocabulary/Tagging.java
+++ b/source/net/yacy/cora/lod/vocabulary/Tagging.java
@@ -163,7 +163,7 @@ public class Tagging {
 	        for (Map.Entry<String, SOTuple> e: table.entrySet()) {
 	            String s = e.getValue() == null ? "" : e.getValue().getSynonymsCSV();
 	            String o = e.getValue() == null ? "" : e.getValue().getObjectlink();
-	            w.write(e.getKey() + (s == null || s.length() == 0 ? "" : ":" + e.getValue()) + (o == null || o.length() == 0 || o.equals(objectspace + e.getKey()) ? "" : "#" + o) + "\n");
+	            w.write(e.getKey() + (s == null || s.length() == 0 ? "" : ":" + e.getValue().getSynonymsCSV()) + (o == null || o.length() == 0 || o.equals(objectspace + e.getKey()) ? "" : "#" + o) + "\n");
 	        }
 	        w.close();
 	        init();