public static void main(String[] args) {
String param1 = args[0].length() > 0 ? args[0] : "brown";
Corpus c = null;
if (param1.equals("brown")) {
c = new Brown();
} else if (param1.equals("negra")) {
c = new Negra();
} else {
System.err.println(
"Illegal parameter! Using standard parameter " + STANDARD_PARAMETER_STRING + " instead.");
param1 = "brown";
c = new Brown();
}
int sizeCorpus = c.getSize();
int sizeSample = (int) (sizeCorpus * CORPUS_PERCENTAGE);
/*
* extracting a sample from the corpus and splitting it up into training set and test set
* (currently only the line numbers, lines themselves are written to file)
*/
c.extractSample(sizeSample);
c.splitSample("tree");
}
// 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 static DepContextCounts importFromReader(BufferedReader in) throws IOException {
Helper.report("[ContextCounts] Importing context word counts...");
DepContextCounts dmc = new DepContextCounts();
String line;
while ((line = in.readLine()) != null) {
if (line.startsWith("<contextcounts")) {
Matcher matcher = contextCountsPattern.matcher(line);
if (matcher.find()) { // ignore first entry: corpus name
Corpus.setName(matcher.group(1));
}
} else if (line.startsWith("<deprelation")) {
Matcher matcher = depRelationPattern.matcher(line);
if (matcher.find()) { // ignore first entry: corpus name
String depRelationString = matcher.group(1);
importDepRelationCounts(in, dmc, depRelationString);
}
} else if (line.equals("</contextcounts>")) {
break;
}
}
Helper.report("[ContextCounts] ...Finished importing context word counts.");
return dmc;
}
public DepContextCounter(String corpusFolderName, int amountOfSentences) {
Corpus.setFolderName(corpusFolderName);
threads = new HashSet<>();
counts = new DepContextCounts();
this.amountOfSentences =
amountOfSentences; // how many sentences per corpus file should be included in the count?
}
@Override
public String toString() {
String s = "CORPUS \"" + Corpus.getName() + "\"\n";
for (String depRelationString : depRelationWordCountMap.keySet()) {
HashMap<String, Long> wordCountMap = depRelationWordCountMap.get(depRelationString);
s += "DEPRELATION \"" + depRelationString + "\"\n";
int i = 0;
for (String contextWord : wordCountMap.keySet()) {
long count = wordCountMap.get(contextWord);
s += count + "\t" + contextWord + "\n";
if (++i >= 5) break;
}
}
return s;
}
