// marginalise over all corpus files using threads
public synchronized void count() {
Helper.report("[ContextCounter] Counting over all corpus files...");
File corpusFolder = new File(DepNeighbourhoodSpace.getProjectFolder(), Corpus.getFolderName());
String[] corpusFilenames = corpusFolder.list();
Arrays.sort(corpusFilenames);
// run each dep marginaliser thread
for (String corpusFilename : corpusFilenames) {
DepContextCounterThread ccThread =
new DepContextCounterThread(
this, corpusFilename, new File(corpusFolder, corpusFilename), amountOfSentences);
threads.add(ccThread);
(new Thread(ccThread)).start();
}
// wait for all threads to finish
try {
while (!threads.isEmpty()) {
wait();
}
} catch (InterruptedException e) {
}
Helper.report("[ContextCounter] ...Finished counting over all corpus files...");
}
public void saveToWriter(BufferedWriter out) throws IOException {
out.write("<contextcounts corpus=\"" + DepNeighbourhoodSpace.getName() + "\">\n");
for (String depRelation : depRelationWordCountMap.keySet()) {
Helper.report(
"[ContextCounts] Writing counts for dep relation \"" + depRelation + "\" to file");
out.write("<deprelation name=\"" + depRelation + "\">\n");
TreeSet<Entry<String, Long>> sortedWordCountMap = getSortedMap(depRelation);
while (!sortedWordCountMap.isEmpty()) {
Entry<String, Long> topWordCountPair = sortedWordCountMap.pollLast();
out.write(topWordCountPair.getValue() + "\t" + topWordCountPair.getKey() + "\n");
}
out.write("</deprelation>\n");
}
out.write("</contextcounts>\n");
}
