/**
* Annotates a file with whatizit
*
* @param outRDF File to be annotated
* @param outputDir Output dor for the annotated RDF (output located at /AO_annotations subfolder)
* @param pmc Article (file) id
* @param pipeline Whatizit pipeline
* @throws Exception
*/
public void annotateFileWithWhatizit(File outRDF, String outputDir, String pipeline)
throws Exception {
logger.info("===Whatizit " + outRDF.getName() + "===");
// 3. Annotate with whatizit
String rdfFileName = outRDF.getAbsolutePath();
String pmc;
if (outRDF.getName().startsWith(this.pmcPrefix)) {
pmc = outRDF.getName().substring(this.pmcPrefix.length(), outRDF.getName().length() - 4);
} else {
pmc = outRDF.getName().substring(0, outRDF.getName().length() - 4);
}
if (pmc.endsWith(this.pmcSuffix)) {
pmc = pmc.substring(0, pmc.length() - this.pmcSuffix.length());
}
String outRDFFileName = outputDir + "/AO_annotations/PMC" + pmc + "_" + pipeline + ".rdf";
List<Annotation> lsAnnotations = new ArrayList<Annotation>();
this.handler.setPaperURLId(GlobalPmc.getPmcRdfUri(pmc));
this.handler.setDocumentPaperId(pmc);
File outRDF_AO = new File(outRDFFileName);
if (!outRDF_AO.exists()) {
outRDF_AO =
this.handler.annotateWithWhatizit(
pipeline,
rdfFileName,
outRDFFileName,
lsAnnotations,
TEXT_PROPERTY,
GlobalPmc.RDF4PMC_AGENT);
}
}
/**
* RDFizes a file
*
* @param subdir File to rdfize
* @param outputDir Output dir for generated RDF
* @param sections Should sections be processed?
* @throws JAXBException
* @throws FileNotFoundException
* @throws UnsupportedEncodingException
* @throws ParserConfigurationException
* @throws SAXException
* @throws PMCIdException
* @throws ArticleTypeException
* @throws DTDException
*/
public File rdfizeFile(File subdir, String outputDir, boolean sections)
throws JAXBException, FileNotFoundException, UnsupportedEncodingException, SAXException,
ParserConfigurationException, DTDException, ArticleTypeException, PMCIdException {
logger.info("===RDFize " + subdir.getName() + "===");
File outRDF = null;
// 1. Create RDF used as a mechanism for improving information retrieval over tagged resources
// as well as to facilitate the discovery of shared conceptualizations[2,3].
this.handler.setStrPmcId(new StringBuilder());
this.handler.setPaper2rdf(new PmcOpenAccess2RDF(subdir, this.handler.getStrPmcId()));
String pmc = this.handler.getStrPmcId().toString();
this.handler.setPaperURLId(GlobalPmc.getPmcRdfUri(pmc));
this.handler.setDocumentPaperId(pmc);
outRDF = new File(outputDir + "/PMC" + pmc + ".rdf");
if (!outRDF.exists()) {
outRDF = this.handler.createRDFFromXML(subdir, outputDir, sections);
}
return outRDF;
}