@Override
protected boolean enhanceMetadata(BxZone zone, DocumentMetadata metadata) {
Matcher simpleMatcher = simplePattern.matcher(zone.toText());
if (simpleMatcher.find()) {
MatchResult result = simpleMatcher.toMatchResult();
String day = result.group(2);
String month = null;
for (int i = 0; i < 12; i++) {
if (result.group(i + 3) != null) {
month = String.valueOf(i + 1);
break;
}
}
String year = result.group(15);
enhanceMetadata(metadata, day, month, year);
return true;
}
Matcher anotherMatcher = anotherPattern.matcher(zone.toText());
if (anotherMatcher.find()) {
MatchResult result = anotherMatcher.toMatchResult();
String day = result.group(14);
String month = null;
for (int i = 0; i < 12; i++) {
if (result.group(i + 2) != null) {
month = String.valueOf(i + 1);
break;
}
}
String year = result.group(15);
enhanceMetadata(metadata, day, month, year);
return true;
}
return false;
}
@Override
public double calculateFeatureValue(BxZone zone, BxPage page) {
return (zone.toText().matches("^.*[\\u0391-\\u03A9].*$")
|| zone.toText().matches("^.*[\\u03B1-\\u03C9].*$"))
? 1
: 0;
}
@Override
public double calculateFeatureValue(BxZone object, BxPage context) {
String text = object.toText();
String[] words = text.split("\\s");
List<Integer> wordLengths = new ArrayList<Integer>(words.length);
for (String word : words) wordLengths.add(word.length());
Collections.sort(wordLengths);
return wordLengths.get(wordLengths.size() / 2);
}
@Override
public double calculateFeatureValue(BxZone zone, BxPage page) {
String[] keywords = {"abstract", "keywords", "key words"
// , "background", "methods", "results", "conclusions", "purpose", "trial", "discussion",
// "summary", "conclusion"
};
for (String keyword : keywords) {
if (zone.toText().toLowerCase().startsWith(keyword)) {
return 1;
}
}
return 0;
}