protected void gogo() throws Exception {
boolean slidingWindow = Configurator.CONFIG.getBoolean("experiments.time.SlidingWindow");
int windowSize = Configurator.CONFIG.getInt("experiments.time.WindowSize");
int maxNumTrgPerSrc = Configurator.CONFIG.getInt("experiments.NumTranslationsToAddPerSource");
String outDir = Configurator.CONFIG.getString("output.Path");
int numThreads = Configurator.CONFIG.getInt("experiments.NumRankingThreads");
boolean doContext = Configurator.CONFIG.getBoolean("experiments.DoContext");
boolean doTime = Configurator.CONFIG.getBoolean("experiments.DoTime");
boolean doEditDist = Configurator.CONFIG.getBoolean("experiments.DoEditDistance");
boolean doAggregate = Configurator.CONFIG.getBoolean("experiments.DoAggregate");
InductPreparer preparer = new InductPreparer();
// Prepare equivalence classes
preparer.prepare();
// Select a subset of src classes to actually induct
Set<EquivalenceClass> srcSubset = preparer.getSrcEqsToInduct();
Set<EquivalenceClass> trgSet = preparer.getTrgEqs();
// Setup scorers
DictScorer contextScorer =
new FungS1Scorer(
preparer.getSeedDict(), preparer.getMaxSrcTokCount(), preparer.getMaxTrgTokCount());
Scorer timeScorer = new TimeDistributionCosineScorer(windowSize, slidingWindow);
Scorer editScorer = new EditDistanceScorer();
// Pre-process properties (i.e. project contexts, normalizes distributions)
preparer.prepareProperties(true, srcSubset, contextScorer, timeScorer);
preparer.prepareProperties(false, preparer.getTrgEqs(), contextScorer, timeScorer);
Collection<EquivClassCandRanking> cands;
Set<Collection<EquivClassCandRanking>> allCands =
new HashSet<Collection<EquivClassCandRanking>>();
if (doTime) {
LOG.info("Ranking candidates using time...");
cands = rank(timeScorer, srcSubset, trgSet, maxNumTrgPerSrc, numThreads);
evaluate(cands, preparer.getSeedDict(), outDir + "time.eval");
EquivClassCandRanking.dumpToFile(preparer.getSeedDict(), cands, outDir + "time.scored");
allCands.add(cands);
}
if (doContext) {
LOG.info("Ranking candidates using context...");
cands = rank(contextScorer, srcSubset, trgSet, maxNumTrgPerSrc, 0.0, numThreads);
evaluate(cands, preparer.getSeedDict(), outDir + "context.eval");
EquivClassCandRanking.dumpToFile(preparer.getSeedDict(), cands, outDir + "context.scored");
allCands.add(cands);
}
if (doEditDist) {
LOG.info("Ranking candidates using edit distance...");
cands = rank(editScorer, srcSubset, trgSet, maxNumTrgPerSrc, numThreads);
evaluate(cands, preparer.getSeedDict(), outDir + "edit.eval");
EquivClassCandRanking.dumpToFile(preparer.getSeedDict(), cands, outDir + "edit.scored");
allCands.add(cands);
}
if (doAggregate) {
LOG.info("Aggregating (MRR) all rankings...");
MRRAggregator aggregator = new MRRAggregator();
cands = aggregator.aggregate(allCands);
evaluate(cands, preparer.getSeedDict(), outDir + "aggmrr.eval");
EquivClassCandRanking.dumpToFile(preparer.getSeedDict(), cands, outDir + "aggmrr.scored");
;
}
LOG.info("--- Done ---");
}
