@Test
public void testLevenshtein() {
Assert.assertEquals(2, StringDistance.levenshtein("GUMBO", "GAMBOL"));
Assert.assertEquals(6, StringDistance.levenshtein("", "GAMBOL"));
Assert.assertEquals(3, StringDistance.levenshtein("Windows", "XP Windows"));
Assert.assertEquals(3, StringDistance.levenshtein("XP Windows", "Windows"));
Assert.assertEquals(0, StringDistance.levenshtein("Windows", "Windows"));
}
@Test
public void testCountSimilar() {
Assert.assertEquals(3, StringDistance.countSimilar("GUMBO", "GAMBOL"));
Assert.assertEquals(0, StringDistance.countSimilar("", "GAMBOL"));
Assert.assertEquals(7, StringDistance.countSimilar("Windows", "XP Windows"));
Assert.assertEquals(7, StringDistance.countSimilar("XP Windows", "Windows"));
Assert.assertEquals(7, StringDistance.countSimilar("Windows", "Windows"));
Assert.assertEquals(3, StringDistance.countSimilar("CMD", "CMD"));
}
/**
* @param object: we want to rerank this
* @param contexts: all contexts
* @param context_table: related contexts to the object
* @return
*/
public Annotation rank(
ArrayList<Annotation> object,
ArrayList<ArrayList<Annotation>> contexts,
Hashtable<String, ArrayList<String>> context_table) {
float max_rank_score = -1;
float final_co_occur = 0;
float final_is_related = 0;
Annotation topRank = new Annotation("", "", 0);
if (object.size() > 0) topRank = object.get(0);
for (int i = 0; i < object.size(); i++) {
Annotation currentAnnotation = object.get(i);
String url1 = currentAnnotation.getAnnotation();
String url1_name = Utils.getDBpediaURLName(url1).toLowerCase();
if (url1_name.startsWith(".")) continue;
// String url_name_for_sim = Utils.getDBpediaURLName(url1).toLowerCase();
String mention = currentAnnotation.getKeyword().replaceAll("\\(.*\\)", "");
float defaultNameSimilarity = (float) computeNameSimilarity(url1, mention, "defaultLabel");
float urlNameSimilarity = (float) JaccardSimilarity.computeSimilarity(url1_name, mention);
float labelNameSimilarity = (float) computeNameSimilarity(url1, mention, "label");
// float nameStringSimilarity = (float)StringDistance.getSim(url1_name,
// currentAnnotation.getKeyword());
// System.out.println(url1+" vs "+currentAnnotation.getKeyword()+": "+defaultNameSimilarity+"
// "+urlNameSimilarity+" "+urlNameSimilarity);
float nameSimilarity = (defaultNameSimilarity + urlNameSimilarity + urlNameSimilarity) / 3;
if (defaultNameSimilarity == 1 || urlNameSimilarity == 1 || labelNameSimilarity == 1) {
nameSimilarity = 1;
}
if (nameSimilarity == 0) {
nameSimilarity = (float) StringDistance.getSim(url1_name, currentAnnotation.getKeyword());
}
System.out.println("processing " + currentAnnotation + " nameSim: " + nameSimilarity);
ArrayList<String> related_context = context_table.get(currentAnnotation.getKeyword());
// System.out.println("related: context:"+related_context);
float co_occur_score = 0;
float is_relate_score = 0;
// System.out.println("all context: "+contexts);
float base = contexts.size() - 1;
for (int j = 0; j < contexts.size(); j++) {
ArrayList<Annotation> currentContext = contexts.get(j);
float max_co_occur = 0;
boolean is_relate_to_the_context = false;
for (int k = 0; k < currentContext.size(); k++) {
if (currentContext.get(k).getKeyword().equals(currentAnnotation.getKeyword())) continue;
if (!related_context.contains(currentContext.get(k).getKeyword())) {
// System.out.println("not relate: "+currentContext.get(k).getKeyword());
continue;
}
String url2 = currentContext.get(k).getAnnotation();
// in case both url1 and url2 are used to annotate some entity mentions: biology, cell
// (Cell_biology)
// if(url1.equals(url2)){
// base--;
// continue;
// }
String url2_name = Utils.getDBpediaURLName(url2).toLowerCase();
float co_occur_count = getRelationCount(url1_name, url2_name);
// float co_occur_count = getRelationCount(url1, url2);
if (!is_relate_to_the_context) {
// is_relate_to_the_context = isURLDirectRelated(url1, url2);
// if(!is_relate_to_the_context){
// is_relate_to_the_context = isURLDirectRelated(url2, url1);
// }
is_relate_to_the_context = isDirectRelated(url1_name, url2_name);
}
// System.out.println("max: "+max_co_occur+" curr: "+co_occur_count);
if (co_occur_count > max_co_occur) {
max_co_occur = co_occur_count;
// System.out.println("max: "+max_co_occur+" curr:
// "+currentContext.get(k).getAnnotation());
}
if (max_co_occur == max_co_occur_threshold && is_relate_to_the_context) {
break;
}
}
if (max_co_occur > 0) {
co_occur_score += (max_co_occur / (double) max_co_occur_threshold);
// currentScore += 1+ (max_co_occur/1000 *
// Math.log(1+max_co_occur_annotation.getScore()));
// System.out.println();
// System.out.println(currentAnnotation.getAnnotation()+" "+max_co_occur+" new score:
// "+co_occur_score);
}
if (is_relate_to_the_context) {
is_relate_score++;
System.out.println("watching is relate score: " + is_relate_score);
}
System.out.println("-------------");
// System.out.println("max_co_occur: "+currentAnnotation.getAnnotation()+"
// "+max_co_occur_annotation.getAnnotation()+" " +currentScore);
}
// currentScore = (float) (1+Math.log(currentScore));
// System.out.println("processing "+currentAnnotation);
// is_relate_score = is_relate_score / contexts.size();
// System.out.println("base: "+contexts.size()+"-1="+base);
// System.out.println("co occur score:
// "+co_occur_score+"^2/"+base+"^2="+((co_occur_score*co_occur_score) / (base * base)));
// System.out.println("is relate score:
// "+is_relate_score+"^2/"+base+"^2="+((is_relate_score*is_relate_score) / (base * base)));
// System.out.println("name score: "+nameSimilarity);
// is_relate_score = (is_relate_score*is_relate_score) / (base * base);
// co_occur_score = (co_occur_score*co_occur_score) / (base * base);
// nameSimilarity = nameSimilarity * nameSimilarity;
float currentScore = 0;
if (base == 0) {
currentScore = (float) (nameSimilarity * Math.log(1 + currentAnnotation.getScore()));
} else {
if (is_relate_score > final_is_related) {
final_is_related = is_relate_score;
}
if (co_occur_score > final_co_occur) {
final_co_occur = co_occur_score;
}
is_relate_score =
(float)
Math.log10(
1
+ (is_relate_score / base)
* 9); // (is_relate_score/ base) * (is_relate_score/ base);
co_occur_score =
(float)
Math.log10(
1
+ (co_occur_score / base)
* 9); // (co_occur_score/ base) * (co_occur_score/ base);
nameSimilarity =
(float) Math.log10(1 + nameSimilarity * 9); // nameSimilarity * nameSimilarity;
// is_relate_score = Math.log(arg0)
System.out.println("co occur score: " + co_occur_score);
System.out.println("is relate score: " + is_relate_score);
System.out.println("name score: " + nameSimilarity);
currentScore =
(float)
(((co_occur_score + nameSimilarity + is_relate_score) / 3)
* Math.log(1 + currentAnnotation.getScore()));
System.out.println("avg: " + ((co_occur_score + nameSimilarity + is_relate_score) / 3));
// currentScore += currentAnnotation.getScore();
}
System.out.println("processing " + currentAnnotation + " -> new: " + currentScore);
if (currentScore > max_rank_score) {
max_rank_score = currentScore;
topRank = currentAnnotation;
System.out.println("top: " + currentAnnotation + " new: " + currentScore);
}
}
topRank.setScore(max_rank_score);
return topRank;
}