use of dkpro.similarity.algorithms.api.SimilarityException in project dkpro-tc by dkpro.
the class SimilarityPairFeatureExtractor method extract.
@Override
public Set<Feature> extract(JCas view1, JCas view2) throws TextClassificationException {
try {
double similarity;
switch(textSimilarityResource.getMode()) {
case text:
similarity = textSimilarityResource.getSimilarity(view1.getDocumentText(), view2.getDocumentText());
break;
case jcas:
similarity = ((JCasTextSimilarityMeasure) textSimilarityResource).getSimilarity(view1, view2);
break;
default:
List<String> f1 = getItems(view1);
List<String> f2 = getItems(view2);
// Remove "_" tokens
for (int i = f1.size() - 1; i >= 0; i--) {
if (f1.get(i) == null || f1.get(i).equals("_")) {
f1.remove(i);
}
}
for (int i = f2.size() - 1; i >= 0; i--) {
if (f2.get(i) == null || f2.get(i).equals("_")) {
f2.remove(i);
}
}
similarity = textSimilarityResource.getSimilarity(f1, f2);
}
return new Feature("Similarity" + textSimilarityResource.getName(), similarity, FeatureType.NUMERIC).asSet();
} catch (FeaturePathException e) {
throw new TextClassificationException(e);
} catch (SimilarityException e) {
throw new TextClassificationException(e);
}
}
use of dkpro.similarity.algorithms.api.SimilarityException in project dkpro-tc by dkpro.
the class CosineFeatureExtractor method extract.
@Override
public Set<Feature> extract(JCas view1, JCas view2) throws TextClassificationException {
try {
TextClassificationTarget aTarget1 = JCasUtil.selectSingle(view1, TextClassificationTarget.class);
TextClassificationTarget aTarget2 = JCasUtil.selectSingle(view2, TextClassificationTarget.class);
// Note: getSimilarity(String, String) is *not* a convenience
// method for getSimilarity(Collection<String>, Collection<String>).
Set<String> text1 = NGramUtils.getDocumentNgrams(view1, aTarget1, true, false, 1, 1, stopwords, ngramAnnotationType).getKeys();
Set<String> text2 = NGramUtils.getDocumentNgrams(view2, aTarget2, true, false, 1, 1, stopwords, ngramAnnotationType).getKeys();
double similarity = measure.getSimilarity(text1, text2);
// Temporary fix for DKPro Similarity Issue 30
if (Double.isNaN(similarity)) {
similarity = 0.0;
}
return new Feature("Similarity" + measure.getName(), similarity, FeatureType.NUMERIC).asSet();
} catch (SimilarityException e) {
throw new TextClassificationException(e);
}
}
Aggregations