private CNGramsStats extractFragment(final String content, String label)
throws SemanticAnalysisException {
int nGramRank = -1;
CText document = new CText(content);
CNGramsStats nGramsFrequencyStats = new CNGramsStats(label);
CNGramsGenerator nGramsExtractor = new CNGramsGenerator(nGramsFrequencyStats);
if (nGramsExtractor.extract(document, content)) {
nGramsFrequencyStats = nGramsExtractor.getNGramsFrequencyStats();
if (nGramsFrequencyStats != null) {
NTaxonomiesConn taxonomyConnection =
CTaxonomyConnectionsPool.getInstance().getLabelsAndCatsConnection();
CTaxonomyExtractor taxonomyExtractor = new CTaxonomyExtractor(taxonomyConnection);
if (taxonomyExtractor.extract(document)) {
/*
* Walk through the list of taxonomyInstance nouns from
* this document document to extract statistics for the label.
*/
int rank = 0;
/*
for( CTaxonomyInstance taxonomyInstance : document.getObjectsMap().values()) {
if( label.compareTo(taxonomyInstance.getLabel()) == 0) {
nGramRank = rank;
break;
}
rank++;
}
*/
}
}
}
return (nGramRank >= 0) ? nGramsFrequencyStats : null;
}
/** Create a text semantic service. */
public CTextSemanticService(short index) throws SemanticAnalysisException {
_taxonomyConnection = CTaxonomyConnectionsPool.getInstance().getLabelsConnection();
_textIndex = index;
}