DO NOT MERGE YET: first pass at instance graph for goals and targets

instutil/Makefile

@@ -0,0 +1,11 @@
+targets.ttl: The-Targets.md
+	./inst2ttl.pl $< > [email protected] && owltools [email protected] && mv [email protected] $@
+.PRECIOUS: targets.ttl
+
+# make like this: remake targets-ann.ttl >& DEBUG
+targets-ann.ttl: targets.ttl
+	blip-findall -debug ann -r gemet -r envo -i ../src/sdgio.obo -goal ix -i $< -consult annotate_goals.pro annotate_label/2 > [email protected]  && mv [email protected] $@
+
+merged.owl: targets.ttl targets-ann.ttl
+	owltools --catalog-xml ../src/catalog-v001.xml $^ ../src/sdgio.owl --merge-support-ontologies --merge-imports-closure -o $@
+

instutil/README.md

@@ -0,0 +1,21 @@
+Experimental approach to automatically extract an OWL representation of the goals and targets.
+
+Opened merged.owl in Protege
+
+The model is
+
+    goal
+      --[has-part]-->
+         target
+           --[has-participant]--> ENVO-continuant
+           --[has-part]--> ENVO-process
+           --[MENTIONS]--> Anything else
+
+As we keep adding to SDGIO, we will keep rerunning the pipeline, and
+in theory the results should improve. We will also add heuristics to
+refine the semantically useless MENTIONS assertions (there are useful
+to indicate that we may be missing something in the ontology).
+
+exclusions can be added directly to the prolog
+
+

instutil/annotate_goals.pro

@@ -0,0 +1,71 @@
+:- use_module(bio(annotator)).
+
+ok('ENVO').
+ok('SDGIO').
+ok('PCO').
+ok('NCBITaxon').
+ok('GEMET').
+not_obo('SDGIO').
+%not_obo('GEMET').
+
+
+ix :-
+        format('@base <http://purl.unep.org/sdg/> .~n'),
+        %format('@prefix ENVO: <http://purl.obolibrary.org/obo/ENVO_> .~n'),
+        format('@prefix SDGIO: <http://purl.unep.org/sdg/SDGIO_> .~n'),
+        format('@prefix has-participant: <http://purl.obolibrary.org/obo/RO_0000057> .~n'),
+        format('@prefix has-part: <http://purl.obolibrary.org/obo/BFO_0000051> .~n'),
+        format('@prefix mentions: <http://purl.unep.org/sdg/MENTIONS> .~n'),
+        forall((ok(S),\+not_obo(S)),
+               format('@prefix ~w: <http://purl.obolibrary.org/obo/~w_> .~n',[S,S])),
+        format('## AUTO:~n'),
+        nl,
+        initialize_annotator.
+
+exclude_list([
+              all,
+              sound
+             ]).
+
+
+annotate_label(C,MC) :-
+        rdfs_label(C,L),
+        exclude_list(XL),
+        sentence_annotate(L,Matches,[excludes([L|XL])]),
+        member(m([MC|_],_,_),Matches),
+        class(MC,MCN),
+        debug(ann,'Annotating: ~w ==> ~w "~w"',[L,MC,MCN]),
+        is_ok(MC),
+        cls_rel(MC,R),
+        format('<~w> ~w ~w .~n',[C,R,MC]),
+        fail.
+
+is_ok(C) :-
+        id_idspace(C,S),
+        ok(S),
+        !.
+is_ok(C) :-
+        debug(ann,'BAD: ~w',[C]),
+        fail.
+
+
+
+cls_rel(C,'has-participant:') :-
+        subclassT(C,'BFO:0000040'),
+        !.
+cls_rel(C,'has-part:') :-
+        subclassT(C,'BFO:0000015'),
+        !.
+cls_rel(_,'mentions:') :- !.
+
+
+
+
+
+/*
+
+blip-findall -goal ix -i z.ttl -consult annotate_goals.pro annotate_label/3
+*/
+
+
+

instutil/inst2ttl.pl

@@ -0,0 +1,69 @@
+#!/usr/bin/perl
+use strict;
+
+my $goal;
+my $tgt;
+
+header();
+
+while(<>) {
+    if (m@^## Goal\s+(\d+)(.*)@) {
+        $goal = ":goal-$1";
+        my $label = trim($2);
+        ttl($goal, 'rdfs:label', lit($label));
+        ttl($goal, 'rdf:type', 'SDGIO:00000000');
+    }
+
+    if (m@^ \*\s+(\d+)\.(\d+)(.*)@) {
+        $tgt = ":target-$1-$2";
+        my $label = trim($3);
+        if ($label =~ m@^by (\d+)(.*)@i) {
+            my $year = ":year-$1";
+            $label = trim($2);
+            ttl($tgt, 'ends-with:', $year);
+            ttl($year, 'rdf:type', 'BFO:0000003');
+        }
+        ttl($tgt, 'rdfs:label', lit($label));
+        ttl($tgt, 'part-of:', $goal);
+        ttl($tgt, 'rdf:type', 'SDGIO:00000001');
+    }
+
+}
+
+exit 0;
+
+sub trim {
+    my $s = shift;
+    $s =~ s@^[\.,\s]+@@g;
+    return $s;
+}
+
+sub lit {
+    my $s = shift;
+    return '"'.$s.'"';
+}
+
+sub ttl {
+    my ($s,$p,$o) = @_;
+    print "$s $p $o .\n";
+}
+
+sub header {
+
+    print <<EOM;
+
+\@base <http://purl.unep.org/sdg/> .
+\@prefix : <http://purl.unep.org/sdg/> .
+\@prefix BFO: <http://purl.obolibrary.org/obo/BFO_> .
+\@prefix part-of: <http://purl.obolibrary.org/obo/BFO_0000050> .
+\@prefix ends-with: <http://purl.obolibrary.org/obo/RO_0002230> .
+\@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
+\@prefix SDGIO: <http://purl.unep.org/sdg/SDGIO_> .
+
+:goals a owl:Ontology .
+
+## Auto-generated
+
+EOM
+
+}