/**
* Create a FeatureVector from the LDA data.
*
* @param c
* @param relevant
* @return
*/
protected UnlabeledFeatureVector<Integer> createUnlabeledVector(Citation c) {
UnlabeledFeatureVector<Integer> fv =
new UnlabeledFeatureVector<Integer>(c.getPmid().toString());
Multimap<Double, Integer> topicDist = lda.getTopics(c.getPmid().toString());
for (double pct : topicDist.keySet()) {
for (int topic : topicDist.get(pct)) {
String str = String.valueOf(topic);
fv.put(str, new DoubleFeature(str, pct));
}
}
return fv;
}
@Override
protected Map<PubmedId, FeatureVector<Integer>> createFeatureVectors(
Collection<Citation> citations) {
StringBuilder sb = new StringBuilder();
for (Citation c : citations) {
sb.append(c.getPmid() + "\tX\t\"" + c.getTitle() + " " + c.getAbstr() + "\"\n");
}
this.lda = new BasicLDA(NUM_TOPICS, ALPHA_SUM_PRIOR, BETA_PRIOR);
try {
lda.train(sb.toString(), NUM_LDA_IT);
} catch (IOException e) {
LOG.error(e);
}
// create the feature vectors
Map<PubmedId, FeatureVector<Integer>> fvs = new HashMap<>();
for (Citation c : citations) {
fvs.put(c.getPmid(), createUnlabeledVector(c));
}
return fvs;
}
