Skip to content
Snippets Groups Projects

Merge branch 'remove-entities-in-plans' into 'master'

Merged Mouhamadou Ba requested to merge cherry-pick-da1f9f6f into master
10 files
+ 179
148
Compare changes
  • Side-by-side
  • Inline
Files
10
+ 54
50
@@ -17,6 +17,10 @@
<param name="input-xslt">
<alias module="read.pubmed" param="xslTransform"/>
</param>
<param name="batch">
<alias module="read.pubmed" param="constantDocumentFeatures"/>
</param>
<param name="outputDir">
<alias module="output.doc-mesh" param="outDir"/>
@@ -139,8 +143,7 @@
<param name="NCBI_taxa_ontobiotope">
<alias module="habitats.tomap-habitats.map-living-organisms" param="mappingFile"/>
</param>
<!-- habitats-->
<param name="ontobiotope-habitat">
<alias module="habitats.tomap-habitats.concept-names" param="oboFiles"/>
@@ -195,14 +198,15 @@
<read>
<pubmed class="XMLReader">
<sourcePath>corpora/&corpus;/batches/&batch;/batch.xml</sourcePath>
<xslTransform>ancillaries/&corpus;-pubmed2alvisnlp.xslt</xslTransform>
<sourcePath>corpora/pubmed/batches/0001/batch.xml</sourcePath>
<xslTransform>ancillaries/microbes-2019-pubmed2alvisnlp.xslt</xslTransform>
<constantDocumentFeatures>batch=0001</constantDocumentFeatures>
</pubmed>
<bionlp-st class="BioNLPSTReader">
<active>true</active>
<sectionName>abstract</sectionName>
<textDir>corpora/&corpus;/batches/&batch;/bionlp-st</textDir>
<textDir>corpora/BioNLP-OST-2019/batches/BB19-kb+ner/bionlp-st</textDir>
</bionlp-st>
</read>
@@ -404,7 +408,7 @@
<!-- Run Yatea term extractor -->
<yatea class="YateaExtractor">
<sectionFilter>@name == "title" or @name == "abstract"</sectionFilter>
<xmlTermsFile>corpora/&corpus;/batches/&batch;/yatea/candidates.xml</xmlTermsFile>
<xmlTermsFile>yatea/candidates.xml</xmlTermsFile>
<posFeature>tt_pos</posFeature>
<configDir>ancillaries/YaTeA/config-habitats</configDir>
<localeDir>ancillaries/YaTeA/locale</localeDir>
@@ -415,7 +419,7 @@
<!-- Run Yatea term extractor on variants -->
<yatea-var class="YateaExtractor">
<sectionFilter>@name == "title" or @name == "abstract"</sectionFilter>
<xmlTermsFile>corpora/&corpus;/batches/&batch;/yatea-var/candidates.xml</xmlTermsFile>
<xmlTermsFile>yatea-var/candidates.xml</xmlTermsFile>
<posFeature>tt_pos</posFeature>
<lemmaFeature>variant</lemmaFeature>
<configDir>ancillaries/YaTeA/config-habitats</configDir>
@@ -669,12 +673,12 @@
<output>
<doc-mesh class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"doc-mesh.txt"</fileName>
<lines>documents.sections:mesh</lines>
<columns separator=";">
"&batch;";
document.@batch;
document.@id;
@UI;
@mesh-name;
@@ -683,12 +687,12 @@
</doc-mesh>
<taxa class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"taxa.txt"</fileName>
<lines>documents.sections.layer:taxa</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
start ^ "-" ^ end;
@@ -702,12 +706,12 @@
</taxa>
<microorganisms class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"microorganisms.txt"</fileName>
<lines>documents.sections.layer:microorganism</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
start ^ "-" ^ end;
@@ -721,12 +725,12 @@
</microorganisms>
<microorganisms-short class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"microorganisms-short.txt"</fileName>
<lines>documents.sections.layer:microorganism[outside:words and not @form == outside:words.@form]</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
start ^ "-" ^ end;
@@ -740,12 +744,12 @@
</microorganisms-short>
<bacteria class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"bacteria.txt"</fileName>
<lines>documents.sections.layer:bacteria</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
start ^ "-" ^ end;
@@ -759,12 +763,12 @@
</bacteria>
<habitats class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"habitats.txt"</fileName>
<lines>documents.sections.layer:habitats</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
start ^ "-" ^ end;
@@ -780,12 +784,12 @@
</habitats>
<phenotypes class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"phenotypes.txt"</fileName>
<lines>documents.sections.layer:phenotypes</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
start ^ "-" ^ end;
@@ -801,12 +805,12 @@
</phenotypes>
<uses class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"uses.txt"</fileName>
<lines>documents.sections.layer:uses</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
start ^ "-" ^ end;
@@ -822,12 +826,12 @@
</uses>
<geo class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"geo.txt"</fileName>
<lines>documents.sections.layer:Geographical</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
start ^ "-" ^ end;
@@ -837,12 +841,12 @@
</geo>
<relations class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"relations.txt"</fileName>
<lines>documents.sections.relations:CooccurrenceLocalization.tuples</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
args:Bacterium.@taxid;
@@ -860,12 +864,12 @@
</relations>
<relations-pheno class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"phenotype-relations.txt"</fileName>
<lines>documents.sections.relations:PhenotypeRelation.tuples</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
args:Microorganism.@taxid;
@@ -883,12 +887,12 @@
</relations-pheno>
<relations-use class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"uses-relations.txt"</fileName>
<lines>documents.sections.relations:UseRelation.tuples</lines>
<columns separator=";">
"&batch;";
document.@batch;
section.document.@id;
section.@name;
args:Microorganism.@taxid;
@@ -974,7 +978,7 @@
</index-sentences>
<sentences class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"sentences.txt"</fileName>
<lines>documents.sections.layer:sentences[@name != "author"]</lines>
@@ -995,7 +999,7 @@
</sentences>
<anaphora class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"anaphora.txt"</fileName>
<lines>documents.sections.relations:coreferences.tuples[args:Ante]</lines>
@@ -1028,7 +1032,7 @@
</anaphora>
<dependencies class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"dependencies.txt"</fileName>
<lines>documents.sections[@name != "author"].relations:dependencies.tuples</lines>
@@ -1247,11 +1251,11 @@
</habitat-ancestors>
<index class="AlvisDBIndexer">
<indexDir>corpora/&corpus;/batches/&batch;/adb</indexDir>
<indexDir>adb</indexDir>
<elements>
<relations>
<items>documents.sections.relations:CooccurrenceLocalization.tuples[args:Bacterium[@bacteria == "true"]]</items>
<id>"&batch;_" ^ id:unique</id>
<id>section.document.@batch ^ id:unique</id>
<name>"Localization"</name>
<type>"localization"</type>
<args>args:Bacterium|args:Localization</args>
@@ -1267,7 +1271,7 @@
</adb>
<index class="AlvisIRIndexer">
<indexDir>corpora/&corpus;/batches/&batch;/index</indexDir>
<indexDir>index</indexDir>
<tokenPositionGap>9216</tokenPositionGap>
<fieldNames>title,abstract,author,full-author,pmid,year,journal,mesh,url</fieldNames>
<relations>
@@ -1404,8 +1408,8 @@
</documents>
</index>
<index-food class="AlvisIRIndexer">
<indexDir>corpora/&corpus;/batches/&batch;/index-food</indexDir>
<!-- <index-food class="AlvisIRIndexer">
<indexDir>.</indexDir>
<tokenPositionGap>9216</tokenPositionGap>
<fieldNames>title,abstract,author,full-author,pmid,year,journal,mesh,url</fieldNames>
<relations>
@@ -1538,7 +1542,7 @@
<keyword>document.@url</keyword>
</fields>
</documents>
</index-food>
</index-food>-->
<!-- HTML visualization -->
<add-feature class="Action">
@@ -1563,14 +1567,14 @@
</add-feature3>
<html class="QuickHTML">
<active>false</active>
<outDir>corpora/&corpus;/batches/&batch;/html</outDir>
<outDir>./html</outDir>
<classFeature>ne-type</classFeature>
<layers>phenotypes,microorganism,habitats</layers>
<colors>#99cc00,#ffcc99,#ffd333,#ffd666</colors>
</html>
<words class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"words.txt"</fileName>
<lines>documents.sections[@name == "title" or @name == "abstract"].layer:words</lines>
@@ -1582,7 +1586,7 @@
<bionlp-st-a2>
<habitats class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;/a2</outDir>
<outDir>a2</outDir>
<files>documents.sections</files>
<fileName>document.@id ^ ".a2"</fileName>
<lines>layer:habitats</lines>
@@ -1595,7 +1599,7 @@
<phenotypes class="TabularExport">
<append/>
<outDir>corpora/&corpus;/batches/&batch;/a2</outDir>
<outDir>a2</outDir>
<files>documents.sections</files>
<fileName>document.@id ^ ".a2"</fileName>
<lines>layer:phenotypes</lines>
@@ -1608,7 +1612,7 @@
<microorganisms class="TabularExport">
<append/>
<outDir>corpora/&corpus;/batches/&batch;/a2</outDir>
<outDir>a2</outDir>
<files>documents.sections</files>
<fileName>document.@id ^ ".a2"</fileName>
<lines>layer:microorganism</lines>
@@ -1621,7 +1625,7 @@
<obt class="TabularExport">
<append/>
<outDir>corpora/&corpus;/batches/&batch;/a2</outDir>
<outDir>a2</outDir>
<files>documents.sections</files>
<fileName>document.@id ^ ".a2"</fileName>
<lines>layer:habitats|layer:phenotypes</lines>
@@ -1633,7 +1637,7 @@
<taxid class="TabularExport">
<append/>
<outDir>corpora/&corpus;/batches/&batch;/a2</outDir>
<outDir>a2</outDir>
<files>documents.sections</files>
<fileName>document.@id ^ ".a2"</fileName>
<lines>layer:microorganism</lines>
@@ -1645,7 +1649,7 @@
<lives-in class="TabularExport">
<append/>
<outDir>corpora/&corpus;/batches/&batch;/a2</outDir>
<outDir>a2</outDir>
<files>documents.sections</files>
<fileName>document.@id ^ ".a2"</fileName>
<lines>relations:CooccurrenceLocalization.tuples[args:Localization.@concept-id != ""]</lines>
@@ -1657,7 +1661,7 @@
<exhibits class="TabularExport">
<append/>
<outDir>corpora/&corpus;/batches/&batch;/a2</outDir>
<outDir>a2</outDir>
<files>documents.sections</files>
<fileName>document.@id ^ ".a2"</fileName>
<lines>relations:PhenotypeRelation.tuples</lines>
@@ -1670,7 +1674,7 @@
<success class="TabularExport">
<outDir>corpora/&corpus;/batches/&batch;</outDir>
<outDir>.</outDir>
<files>$</files>
<fileName>"success.txt"</fileName>
<lines>documents</lines>
Loading