Merge pull request #61 from HHS/meshrdf-dev

Changes for MeSH 2016
HHS · Nov 19, 2015 · f36bfe3 · f36bfe3
2 parents 0dccb57 + cf68e98
commit f36bfe3
Show file tree

Hide file tree

Showing 5 changed files with 126 additions and 25 deletions.
diff --git a/bin/fetch-mesh-xml.sh b/bin/fetch-mesh-xml.sh
@@ -1,4 +1,15 @@
-#!/bin/sh -e
+#!/bin/bash 
+
+
+while getopts "h:y:u:" opt; do
+    case $opt in 
+        h) export MESHRDF_HOME=$OPTARG ;;
+        y) export MESHRDF_YEAR=$OPTARG ;;
+        u) export MESHRDF_URI=$OPTARG ;;
+        *) echo "Usage: $0 [-h meshrdf-home] [-y year] [-u uri] [-i]" 1>&2 ; exit 1 ;;
+    esac
+done
+shift $(($OPTIND - 1))
 
 if [ -z "$MESHRDF_HOME" ]; then
     echo "Please define MESHRDF_HOME environment variable" 1>&2
@@ -8,18 +19,18 @@ fi
 mkdir -p "$MESHRDF_HOME/data"
 
 # CAn override default year with MESHRDF_YEAR environment variable
-YEAR=${MESHRDF_YEAR:-2015}
+YEAR=${MESHRDF_YEAR:-2016}
 
 # Can override default URI with MESHRDF_URI environment variable
 URI=${MESHRDF_URI:-ftp://ftp.nlm.nih.gov/online/mesh/$YEAR}
 
-wget "$URI/desc$YEAR.dtd" -O "$MESHRDF_HOME/data/desc$YEAR.dtd"
 wget "$URI/desc$YEAR.xml" -O "$MESHRDF_HOME/data/desc$YEAR.xml"
-wget "$URI/pa$YEAR.dtd" -O "$MESHRDF_HOME/data/pa$YEAR.dtd"
-wget "$URI/pa$YEAR.xml" -O "$MESHRDF_HOME/data/pa$YEAR.xml"
-wget "$URI/qual$YEAR.dtd" -O "$MESHRDF_HOME/data/qual$YEAR.dtd"
 wget "$URI/qual$YEAR.xml" -O "$MESHRDF_HOME/data/qual$YEAR.xml"
-wget "$URI/supp$YEAR.dtd" -O "$MESHRDF_HOME/data/supp$YEAR.dtd"
 wget "$URI/supp$YEAR.xml" -O "$MESHRDF_HOME/data/supp$YEAR.xml"
 
+if [ $YEAR -le 2015 ]; then
+    wget "$URI/desc$YEAR.dtd" -O "$MESHRDF_HOME/data/desc$YEAR.dtd"
+    wget "$URI/qual$YEAR.dtd" -O "$MESHRDF_HOME/data/qual$YEAR.dtd"
+    wget "$URI/supp$YEAR.dtd" -O "$MESHRDF_HOME/data/supp$YEAR.dtd"
+fi
 
diff --git a/bin/mesh-xml2rdf.sh b/bin/mesh-xml2rdf.sh
@@ -30,6 +30,17 @@ else
 fi
 cd $($READLINK -e `dirname $0`/..)
 
+while getopts "h:j:y:u" opt; do
+  case $opt in
+    h) export MESHRDF_HOME=$OPTARG ;;
+    j) export SAXON_JAR=$OPTARG ;;
+    y) export MESHRDF_YEAR=$OPTARG ;; 
+    u) export MESHRDF_URI_YEAR="yes" ;;
+    *) echo "Usage: $0 [-h mesh-rdf-home] [-j saxon-jar-path] [-y year ]" 1>&2 ; exit 1 ;;
+  esac
+done
+shift $(($OPTIND - 1))
+
 # Check for some needed environment variables
 if [ -z "$MESHRDF_HOME" ]; then
     echo "Please define MESHRDF_HOME environment variable" 1>&2
@@ -42,7 +53,7 @@ if [ -z "$SAXON_JAR" ]; then
 fi
 
 # Can override default year with MESHRDF_YEAR environment variable
-YEAR=${MESHRDF_YEAR:-2015}
+YEAR=${MESHRDF_YEAR:-2016}
 
 # Set the output file name, and the parameter that controls the RDF URIs,
 # according to whether or not MESHRDF_URI_YEAR is "yes"

diff --git a/meta/vocabulary.ttl b/meta/vocabulary.ttl
@@ -9,7 +9,7 @@
 
 <http://id.nlm.nih.gov/mesh/vocab#> rdf:type :Ontology ;
 
-                                    :versionInfo "0.9" .
+                                    :versionInfo "0.9.3" .
 
 
 #################################################################
@@ -304,6 +304,18 @@ dct:description rdf:type :AnnotationProperty .
 
 
 
+###  http://id.nlm.nih.gov/mesh/vocab#active
+
+<http://id.nlm.nih.gov/mesh/vocab#active> rdf:type :DatatypeProperty ;
+
+                                          rdfs:label "active" ;
+
+                                          dct:description "A property of MeSH objects indicating whether they still appear in the current version of the most recently released MeSH year" ;
+
+                                          rdfs:range xsd:boolean .
+
+
+
 ###  http://id.nlm.nih.gov/mesh/vocab#altLabel
 
 <http://id.nlm.nih.gov/mesh/vocab#altLabel> rdf:type :DatatypeProperty ;
@@ -418,6 +430,18 @@ dct:description rdf:type :AnnotationProperty .
 
 
 
+###  http://id.nlm.nih.gov/mesh/vocab#lastActive
+
+<http://id.nlm.nih.gov/mesh/vocab#lastActiveYear> rdf:type :DatatypeProperty ;
+
+                                              rdfs:label "lastActiveYear" ;
+
+                                              dct:description "The lastActiveYear property value is the year in which the subject last appeared.  If that year is still hosted, a new URI can be constructed." ;
+
+                                              rdfs:range xsd:string .
+
+
+
 ###  http://id.nlm.nih.gov/mesh/vocab#lexicalTag
 
 <http://id.nlm.nih.gov/mesh/vocab#lexicalTag> rdf:type :DatatypeProperty ;
@@ -428,6 +452,18 @@ dct:description rdf:type :AnnotationProperty .
 
 
 
+###  http://id.nlm.nih.gov/mesh/vocab#nlmClassificationNumber
+
+<http://id.nlm.nih.gov/mesh/vocab#nlmClassificationNumber> rdf:type :DatatypeProperty ;
+
+                                                           rdfs:label "nlmClassificationNumber" ;
+
+                                                           dct:description "Each MeSH Descriptor has a corresponding class number in the NLM Classification.  This classification is similar to the Library of Congress Classification (LCC).";
+
+                                                           rdfs:range xsd:string .
+
+
+
 ###  http://id.nlm.nih.gov/mesh/vocab#note
 
 <http://id.nlm.nih.gov/mesh/vocab#note> rdf:type :DatatypeProperty ;

diff --git a/meta/void.ttl b/meta/void.ttl
@@ -15,7 +15,7 @@
     dcterms:creator "U.S. National Library of Medicine" ;
     dcterms:date "2014-11-01"^^xsd:date ;
     foaf:primaryTopic <http://id.nlm.nih.gov/mesh/void#MeSHRDF> ;
-    owl:versionInfo "0.9" .
+    owl:versionInfo "0.9.3" .
 
 
 # Metadata description of the Dataset
@@ -57,15 +57,7 @@
     void:exampleResource <http://id.nlm.nih.gov/mesh/D000001> ;
 
 # FTP dump file
-    void:dataDump "ftp://ftp.nlm.nih.gov/online/mesh/2014/desc2014.nt" ;
-    void:dataDump "ftp://ftp.nlm.nih.gov/online/mesh/2014/qual2014.nt" ;
-    void:dataDump "ftp://ftp.nlm.nih.gov/online/mesh/2014/supp2014.nt" ;
+    void:dataDump "ftp://ftp.nlm.nih.gov/online/mesh/mesh.nt" ;
+    void:dataDump "ftp://ftp.nlm.nih.gov/online/mesh/2016/mesh2016.nt" ;
+    void:dataDump "ftp://ftp.nlm.nih.gov/online/mesh/2015/mesh2015.nt" .
 
-# triple count
-    void:triples "16644674"^^xsd:long ;
-
-# entity count
-    void:entities "1925662"^^xsd:long ;
-
-# distinct subject
-    void:distinctSubjects "1925662"^^xsd:long .
diff --git a/xslt/desc.xsl b/xslt/desc.xsl
@@ -18,6 +18,7 @@
     <xsl:for-each select="DescriptorRecordSet/DescriptorRecord">
 
       <xsl:variable name='descriptor_id' select='DescriptorUI'/>
+      <xsl:variable name='descriptor_label' select='DescriptorName/String'/>
 
       <!--
         Transformation rule: meshv:identifier
@@ -27,7 +28,7 @@
           <xsl:value-of select="$descriptor_id"/>
         </uri>
       </xsl:variable>
-      
+
       <xsl:call-template name='triple'>
         <xsl:with-param name="doc">
           <desc>This relation states that a descriptor record has a unique identifier.</desc>
@@ -82,7 +83,7 @@
           <xsl:copy-of select="$descriptor_uri"/>
           <uri prefix='&rdfs;'>label</uri>
           <literal lang='en'>
-            <xsl:value-of select="DescriptorName/String"/>
+            <xsl:value-of select='$descriptor_label'/>
           </literal>
         </xsl:with-param>
       </xsl:call-template>
@@ -129,6 +130,22 @@
           </xsl:with-param>
         </xsl:call-template>
 
+        <!--
+          Give it a label
+        -->
+        <xsl:call-template name='triple'>
+          <xsl:with-param name="doc">
+            <desc>Provide a name for the AllowedDescriptorQualifierPair</desc>
+          </xsl:with-param>
+          <xsl:with-param name="spec">
+            <xsl:copy-of select='$dqpair_uri'/>
+            <uri prefix='&rdfs;'>label</uri>
+            <literal lang='en'>
+              <xsl:value-of select="concat($descriptor_label, '/', QualifierReferredTo/QualifierName/String)"/>
+            </literal>
+          </xsl:with-param>
+        </xsl:call-template>
+
         <!-- 
           Link it back to the descriptor
         -->
@@ -157,7 +174,25 @@
           </xsl:with-param>
         </xsl:call-template>        
       </xsl:for-each>
-
+
+      <!--
+        NLMClassificationNumber
+      -->
+      <xsl:if test="NLMClassificationNumber">
+        <xsl:call-template name='triple'>
+          <xsl:with-param name="doc">
+            <desc>Each descriptor may have an NLMClassificationNumber.</desc>
+          </xsl:with-param>
+          <xsl:with-param name='spec'>
+            <xsl:copy-of select="$descriptor_uri"/>
+            <uri prefix='&meshv;'>nlmClassificationNumber</uri>
+            <literal>
+              <xsl:value-of select="NLMClassificationNumber"/>
+            </literal>
+          </xsl:with-param>
+        </xsl:call-template>
+      </xsl:if>
+
       <!--
         Transformation rule: publicMeSHNote
       -->
@@ -176,7 +211,7 @@
           </xsl:with-param>
         </xsl:call-template>
       </xsl:if>
-      
+
       <xsl:for-each select="EntryCombinationList/EntryCombination">
 
         <xsl:variable name='ecin_uri'>
@@ -200,6 +235,22 @@
           </xsl:with-param>
         </xsl:call-template>
 
+        <!-- 
+          Give it a label
+        -->
+        <xsl:call-template name='triple'>
+          <xsl:with-param name="doc">
+            <desc>Provide a label for the DisallowedDescriptorQualifierPair.</desc>
+          </xsl:with-param>
+          <xsl:with-param name="spec">
+            <xsl:copy-of select='$ecin_uri'/>
+            <uri prefix='&rdfs;'>label</uri>
+            <literal lang='en'>
+              <xsl:value-of select="concat(ECIN/DescriptorReferredTo/DescriptorName/String,
+                                           '/', ECIN/QualifierReferredTo/QualifierName/String)"/>
+            </literal>
+          </xsl:with-param>
+        </xsl:call-template>
         <!-- 
           Link it back to the descriptor
         -->