public List<String[]> getRankedEntities(String entity, List<String> contextPhrases) {
List<String[]> rankedEntities = new ArrayList<String[]>();
PriorityQueue<String[]> queue = new PriorityQueue<String[]>();
contextPhrases.remove(entity);
StringBuffer contextQuery = new StringBuffer();
for (String c : contextPhrases) {
contextQuery.append("\"" + c + "\"" + " ");
}
int contextCount = YahooBOSS.makeQuery(contextQuery.toString());
String xml = "";
if ((xml = Wikiminer.getXML(entity, false)) != null) {
ArrayList<String[]> senses = Wikiminer.getWikipediaSenses(xml, true);
for (String[] senseArr : senses) {
int senseCount = YahooBOSS.makeQuery('"' + senseArr[0] + "\" " + contextQuery.toString());
queue.add(senseArr, ((double) senseCount / (double) contextCount));
}
}
while (queue.hasNext()) {
rankedEntities.add(queue.next());
}
return rankedEntities;
}
public static void main(String[] args) {
ExtractEntities ee = new ExtractEntities();
DocumentBuilder db = null;
DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance();
BufferedWriter bw = null;
BufferedReader br = null;
try {
br = new BufferedReader(new FileReader("data/" + args[0] + ".txt"));
bw = new BufferedWriter(new FileWriter("data/" + args[0] + "_wiki_entities.txt"));
} catch (Exception e) {
e.printStackTrace();
}
try {
db = dbf.newDocumentBuilder();
} catch (Exception e) {
e.printStackTrace();
}
String line = "";
try {
while ((line = br.readLine()) != null) {
bw.write("=====================================================\n");
bw.write("Tweet: " + line + "\n");
bw.write("=====================================================\n");
HashSet<String> entities = ee.getEntitiesinTweet(line);
List<String> contextPhrases = new ArrayList<String>(entities);
for (String entity : entities) {
bw.write("query: " + entity + "\n");
List<String[]> rankedEntities = ee.getRankedEntities(entity, contextPhrases);
// bw.write("RankedEntities: "+rankedEntities+"\n");
for (String[] entityArr : rankedEntities) {
bw.write("Entity: " + entityArr[0] + "\n");
String xml = Wikiminer.getXML(entityArr[1], true);
bw.write(
"RankedTypes: "
+ Wikiminer.getRankedTypes(entityArr[0], xml, contextPhrases, 5)
+ "\n");
}
bw.write("--------------------------------------------------\n");
}
bw.write("\n");
}
bw.flush();
} catch (Exception e) {
e.printStackTrace();
}
}
