update of evaluation

DH-RSE · Feb 22, 2024 · 07db30a · 07db30a
1 parent 22ef712
commit 07db30a
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 41 deletions.
diff --git a/evaluation/csv/citation-types-frequencies.csv b/evaluation/csv/citation-types-frequencies.csv
@@ -1,8 +1,8 @@
-Citation type, abs. frequency (n=119), rel. frequency (in %),ALL / abs. frequency (n=119),ALL / rel. frequency (in %)
-Soft.Bib,1,.84,1,.84
-Soft.Bib.Ref,0,.0,0,.0
-Soft.Name,119,100.0,119,100.0
-Soft.Agent,0,.0,0,.0
-Soft.URL,9,7.56,9,7.56
-Soft.PID,0,.0,0,.0
-Soft.Ver,0,.0,0,.0
+Citation type, abs. frequency (n=234), rel. frequency (in %),ALL / abs. frequency (n=234),ALL / rel. frequency (in %)
+soft.bib,3,1.28,3,1.28
+soft.bib.ref,27,11.54,27,11.54
+soft.name,231,98.72,231,98.72
+soft.agent,35,14.96,35,14.96
+soft.url,86,36.75,86,36.75
+soft.pid,0,.0,0,.0
+soft.ver,4,1.71,4,1.71
diff --git a/evaluation/xsl/citation-types-frequencies.xsl b/evaluation/xsl/citation-types-frequencies.xsl
@@ -12,55 +12,33 @@
     <xsl:output method="text" omit-xml-declaration="yes"/>
 
 
+    <xsl:include href="global-parameters.xsl"/>
+
+
     <!-- Keys -->    
     <xsl:key name="rs-by-ref" match="*:rs" use="@ref"/>
     <xsl:key name="ptr-by-target" match="*:ptr" use="@target"/>
 
-    <!-- Global variables -->
-
-    <!-- directories with TEI files -->
-    <xsl:variable name="collection-dirs" as="xs:string+" select="(  
-        '../../data/JTEI/10_2016-19',
-        '../../data/JTEI/13_2020-22',
-        '../../data/JTEI/7_2014',
-        '../../data/JTEI/rolling_2019',
-        '../../data/JTEI/rolling_2023',
-        '../../data/JTEI/11_2019-20',
-        '../../data/JTEI/12_2019-20',
-        '../../data/JTEI/14_2021-23',
-        '../../data/JTEI/8_2014-15',
-        '../../data/JTEI/rolling_2021',
-        '../../data/JTEI/16_2023_spa',
-        '../../data/JTEI/9_2016-17',
-        '../../data/JTEI/rolling_2022'
-        )"/>
-
-    <!-- software types -->
-    <xsl:variable name="types" select="document('../../taxonomy/citation-taxonomy.xml')//*:taxonomy/*:category/@xml:id" as="xs:string+"/>
+    <!-- Global variables -->    
+
 
     <!-- additional columns for predefined filters (applied to file path, compare directory list above for examples) -->
     <!-- could also be ('/deu/', '/eng/', ...) for instance -->
     <!-- could also be ('/2015/', '/2016/', '/2017/', '/2018/', '/2019/', '/2020/') for instance -->
     <xsl:variable name="path-filters" select="''" as="xs:string+"/>
 
-    <!-- character to be used as CSV separator -->
-    <xsl:variable name="csv-separator" select="','" as="xs:string"/>
-
-    <!-- newline character -->
-    <xsl:variable name="NEWLINE"><xsl:text>
-</xsl:text></xsl:variable>
 
     <!-- instances of one specific software in one text -->
     <xsl:variable name="instances">
-        <xsl:for-each select="$collection-dirs">            
+        <xsl:for-each select="$COLLECTION-DIRS">            
             <xsl:for-each select="collection(concat(., '?select=*.xml;recurse=yes;on-error=warning'))">
                 <xsl:variable name="path" select="base-uri()" as="xs:string"/>
                 <xsl:variable name="doc" select="/"/>
                 <xsl:for-each select="distinct-values($doc//*:ptr[@type='software']/@target)">                
                     <xsl:variable name="current-software-target" select="." as="xs:string"/>
                     <xsl:variable name="rs-for-this-software" select="key('rs-by-ref', key('ptr-by-target', $current-software-target, $doc)/concat('#', @xml:id), $doc)"/>
                     <instance text="{$path}" software="{$current-software-target}">
-                        <xsl:for-each select="$types">
+                        <xsl:for-each select="$SOFTWARE-MENTION-TYPES">
                             <!-- counting frequencies -->
                             <!--<xsl:attribute name="{.}" select="count($rs-with-this-key[contains(lower-case(@ana), lower-case(concat('#',current())))])"/>-->
                             <!-- determining presence -->
@@ -91,7 +69,7 @@
         <xsl:value-of select="$NEWLINE"/>
 
 
-        <xsl:for-each select="$types">
+        <xsl:for-each select="$SOFTWARE-MENTION-TYPES">
             <xsl:variable name="citation-type" select="." as="xs:string"/>
 
             <xsl:value-of select="concat($citation-type, ',')"/>

diff --git a/evaluation/xsl/citation-types.xsl b/evaluation/xsl/citation-types.xsl
@@ -51,11 +51,11 @@
 
                 <xsl:for-each select="distinct-values($doc//*:rs[some $cit in $categories[position() &gt; 2] satisfies contains(@type, $cit)]/tokenize(@ref, ' '))">
                     <xsl:sort select="." order="ascending"/>
-                    <xsl:variable name="current-ref" select="substring-after(., '#')" as="xs:string"/>
-                    <xsl:variable name="software-id" select="lower-case(substring-after(key('software-target-by-id', $current-ref, $doc)[1], '#'))"/>
+                    <xsl:variable name="current-ref" select="." as="xs:string"/>
+                    <xsl:variable name="software-id" select="lower-case(substring-after(key('software-target-by-id', substring-after($current-ref, '#'), $doc)[1], '#'))"/>
                     <xsl:variable name="software-name" select="(
                         key('software-name-by-id', $software-id, $software-list), 
-                        error((), concat('No software name found for ID ''', $software-id, ''' / ref ''#', $current-ref ,''' (', base-uri($doc), ')')))[1]" as="xs:string"/>
+                        error((), concat('No software name found for ID ''', $software-id, ''' / ref ''', $current-ref ,''' (', base-uri($doc), ')')))[1]" as="xs:string"/>
                     <xsl:variable name="rs-with-this-ref" select="key('rs-by-ref', ., $doc)"/>