private static void transitive_reductionInstance(GAction action, G g) throws SLIB_Ex_Critic {
int invalidInstanceNb = 0;
int annotNbBase = 0;
int annotDeleted = 0;
Set<URI> instances = GraphAccessor.getInstances(g);
logger.info("Cleaning " + RDF.TYPE + " triplets of " + g.getURI());
System.out.println(g);
RVF_TAX rvf = new RVF_TAX(g, Direction.IN);
// Retrieve descendants for all vertices
Map<URI, Set<URI>> descs = rvf.getAllRVClass();
for (URI instance : instances) {
Set<URI> redundants = new HashSet<URI>();
Set<URI> classes = g.getV(instance, RDF.TYPE, Direction.OUT);
annotNbBase += classes.size();
Iterator<URI> it = classes.iterator();
while (it.hasNext()) {
URI c = it.next();
Set<URI> descC = descs.get(c);
for (URI c2 : classes) {
if (c != c2 && descC.contains(c2)) {
redundants.add(c);
it.remove();
break;
}
}
}
if (!redundants.isEmpty()) {
for (URI r : redundants) {
g.removeE(new Edge(instance, RDF.TYPE, r));
}
invalidInstanceNb++;
annotDeleted += redundants.size();
}
}
double invalidInstanceP = 0;
if (instances.size() > 0) {
invalidInstanceP = invalidInstanceNb * 100 / instances.size();
}
double annotDelP = 0;
if (annotNbBase > 0) {
annotDelP = annotDeleted * 100 / annotNbBase;
}
logger.info(
"Number of instance containing abnormal annotation: "
+ invalidInstanceNb
+ "/"
+ instances.size()
+ " i.e. ("
+ invalidInstanceP
+ "%)");
logger.info(
"Number of annotations: "
+ annotNbBase
+ ", deleted: "
+ annotDeleted
+ " ("
+ (annotDelP)
+ "%), current annotation number "
+ (annotNbBase - annotDeleted));
}
/**
* Reduction of the set of vertices composing the graph.
*
* <p>------------------------------------------------------------------- IMPORTANT: If modified,
* this documentation must also be modified in the class GActionType.
* -------------------------------------------------------------------
*
* <p>Accepted parameters are:
*
* <ul>
* <li>regex: specify a REGEX in Java syntax which will be used to test if the value associated
* to a vertex makes it eligible to be removed. If the value match the REGEX, the vertex
* will be removed
* <li>vocabulary: Remove all the vertices associated to the vocabularies specified. Accepted
* vocabularies flag are RDF, RDFS, OWL. Several vocabularies can be specified using comma
* separator.
* <li>file_uris: specify a list of files containing URIs corresponding to the vertices to
* remove. Multiple files can be specified using comma separator.
* </ul>
*
* @param factory the factory to consider if element requires to be generated (e.g. {@link URI})
* @param action the action to perform
* @param g the graph on which the action must be performed
* @throws SLIB_Ex_Critic
*/
private static void verticeReduction(URIFactory factory, GAction action, G g)
throws SLIB_Ex_Critic {
logger.info("-------------------------------------");
logger.info(" Vertices Reduction");
logger.info("-------------------------------------");
logger.info("Starting " + GActionType.VERTICES_REDUCTION);
String regex = (String) action.getParameter("regex");
String vocVal = (String) action.getParameter("vocabulary");
String file_uris = (String) action.getParameter("file_uris");
String rootURIs = (String) action.getParameter("root_uri");
Set<URI> classes = GraphAccessor.getClasses(g);
Set<URI> instances = GraphAccessor.getInstances(g);
logger.info("Classes : " + classes.size());
logger.info("instances: " + instances.size());
logger.info("vertices : " + g.getV().size());
Set<URI> toRemove = new HashSet<URI>();
if (rootURIs != null) {
/*
* Reduce the Graph considering all classes subsumed by the given root vertex
* Instances annotated by those classes are also conserved into the graph, others are removed.
*/
logger.info(
"Applying reduction of the part of the graph "
+ g.getURI()
+ " which is not contained in the graph induced by "
+ rootURIs
+ " (only the classes subsumed by the given root are considered)");
try {
URI rootURI = factory.getURI(rootURIs);
if (!g.containsVertex(rootURI)) {
throw new SLIB_Ex_Critic(
"Error cannot state vertex associated to URI " + rootURI + " in graph " + g.getURI());
}
DescendantEngine descEngine = new DescendantEngine(g);
Set<URI> descsInclusive = descEngine.getDescendantsInc(rootURI);
logger.info(descsInclusive.size() + " subclasses of " + rootURI + " detected");
int classesNb = classes.size();
Set<URI> classesToRemove = classes;
classesToRemove.removeAll(descsInclusive);
logger.info(
"Removing " + classesToRemove.size() + "/" + classesNb + " classes of the graph");
g.removeV(classesToRemove);
// We then remove the entities which are not
// linked to the graph current underlying taxonomic graph
Set<URI> instancesToRemove = new HashSet<URI>();
for (URI v : instances) {
// No links to taxonomic graph anymore
// we check the URI as is not considered as both instance and class
if (!descsInclusive.contains(v) && g.getV(v, RDF.TYPE, Direction.OUT).isEmpty()) {
instancesToRemove.add(v);
}
}
logger.info("Removing " + instancesToRemove.size() + " instances of the graph");
g.removeV(instancesToRemove);
} catch (IllegalArgumentException e) {
throw new SLIB_Ex_Critic(
"Error value specified for parameter root_uri, i.e. "
+ rootURIs
+ " cannot be converted into an URI");
}
} else if (regex != null) {
logger.info("Applying regex: " + regex);
Pattern pattern;
try {
pattern = Pattern.compile(regex);
} catch (PatternSyntaxException e) {
throw new SLIB_Ex_Critic(
"The specified regex '" + regex + "' is invalid: " + e.getMessage());
}
Matcher matcher;
for (URI v : g.getV()) {
matcher = pattern.matcher(v.stringValue());
if (matcher.find()) {
toRemove.add(v);
logger.debug("regex matches: " + v);
}
}
logger.info("Vertices to remove: " + toRemove.size() + "/" + g.getV().size());
g.removeV(toRemove);
logger.debug("ending " + GActionType.VERTICES_REDUCTION);
} else if (vocVal != null) {
String[] vocs = vocVal.split(",");
for (String voc : vocs) {
if (voc.trim().equals("RDF")) {
logger.info("Removing RDF vocabulary");
removeVocURIs(factory, getRDFVocURIs(), g);
} else if (voc.trim().equals("RDFS")) {
logger.info("Removing RDFS vocabulary");
removeVocURIs(factory, getRDFSVocURIs(), g);
} else if (voc.trim().equals("OWL")) {
logger.info("Removing OWL vocabulary");
removeVocURIs(factory, getOWLVocURIs(), g);
}
}
} else if (file_uris != null) {
String[] files = file_uris.split(",");
for (String f : files) {
logger.info("Removing Uris specified in " + f);
try {
FileInputStream fstream = new FileInputStream(f.trim());
DataInputStream in = new DataInputStream(fstream);
BufferedReader br = new BufferedReader(new InputStreamReader(in));
String line;
while ((line = br.readLine()) != null) {
line = line.trim();
g.removeV(factory.getURI(line));
}
in.close();
} catch (IOException e) {
throw new SLIB_Ex_Critic(e.getMessage());
}
}
}
logger.info("vertices reduction performed");
logger.info("-------------------------------------");
}
