public static void main(String[] args) throws IOException {
String workDir = "E:/dev_workspace/tmp/workspace/duc2007";
String idfFilename = "duc2007.idf";
final double TOTAL_PAGE_COUNT = 30000000000.0D;
Map<String, Double> idfValues = new HashMap<String, Double>();
File idfFIle = FileUtils.getFile(workDir + "/" + DIR_IDF_FILE, idfFilename);
log.info("Loading idf value file[" + idfFIle.getAbsolutePath() + "]");
LineIterator lineIterator = null;
try {
lineIterator = FileUtils.lineIterator(idfFIle, DEFAULT_CHARSET.toString());
while (lineIterator.hasNext()) {
String line = lineIterator.nextLine();
String[] strs = line.split("###");
if (strs.length != 2) {
log.warn("Line[" + line + "] format is illegal, ignore it!");
continue;
}
idfValues.put(strs[0].trim(), Long.parseLong(strs[1]) / TOTAL_PAGE_COUNT);
}
log.info("Load idf value file[" + idfFIle.getAbsolutePath() + "] finished!");
} catch (IOException e) {
log.error("Load idf value file[" + idfFIle.getAbsolutePath() + "] error!", e);
throw e;
} finally {
if (lineIterator != null) {
lineIterator.close();
}
}
String question =
"Describe the legal battle between various recording artists and members of the record industry and the Internet music site Napster. What support, or lack thereof, have the litigants received?";
EhCacheUtil ehCacheUtil = new EhCacheUtil("db_cache_vec", "lab");
SummaryBuilderByVector summaryBuilder =
new SummaryBuilderByVector(
workDir, "0", "D0714D.txt", 10, idfValues, question, ehCacheUtil, 1.0f, 1.6f);
ExecutorService es = Executors.newSingleThreadExecutor();
Future<Boolean> future = es.submit(summaryBuilder);
try {
future.get();
} catch (InterruptedException | ExecutionException e) {
e.printStackTrace();
}
es.shutdown();
EhCacheUtil.close();
}
public WriterOutputStream(Writer out) {
this.writer = out;
decoder = DEFAULT_CHARSET.newDecoder();
decoder.onMalformedInput(CodingErrorAction.REPLACE);
decoder.onUnmappableCharacter(CodingErrorAction.REPLACE);
}