protected internal static void GetRelevantWordsBase(Token t, CollectionValuedMap <string, string> relWords)
{
if (t != null)
{
IDictionary <string, string> str = t.ClassORRestrictionsAsString();
if (str != null)
{
relWords.AddAll(str);
}
}
}
private void RunParallelApplyPats(IDictionary <string, DataInstance> sents, string label, E pattern, TwoDimensionalCounter <CandidatePhrase, E> wordsandLemmaPatExtracted, CollectionValuedMap <E, Triple <string, int, int> > matchedTokensByPat, ICollection
<CandidatePhrase> alreadyLabeledWords)
{
Redwood.Log(Redwood.Dbg, "Applying pattern " + pattern + " to a total of " + sents.Count + " sentences ");
IList <string> notAllowedClasses = new List <string>();
IList <string> sentids = CollectionUtils.ToList(sents.Keys);
if (constVars.doNotExtractPhraseAnyWordLabeledOtherClass)
{
foreach (string l in constVars.GetAnswerClass().Keys)
{
if (!l.Equals(label))
{
notAllowedClasses.Add(l);
}
}
notAllowedClasses.Add("OTHERSEM");
}
IDictionary <TokenSequencePattern, E> surfacePatternsLearnedThisIterConverted = null;
IDictionary <SemgrexPattern, E> depPatternsLearnedThisIterConverted = null;
if (constVars.patternType.Equals(PatternFactory.PatternType.Surface))
{
surfacePatternsLearnedThisIterConverted = new Dictionary <TokenSequencePattern, E>();
string patternStr = null;
try
{
patternStr = pattern.ToString(notAllowedClasses);
TokenSequencePattern pat = ((TokenSequencePattern)TokenSequencePattern.Compile(constVars.env[label], patternStr));
surfacePatternsLearnedThisIterConverted[pat] = pattern;
}
catch (Exception e)
{
log.Info("Error applying pattern " + patternStr + ". Probably an ill formed pattern (can be because of special symbols in label names). Contact the software developer.");
throw;
}
}
else
{
if (constVars.patternType.Equals(PatternFactory.PatternType.Dep))
{
depPatternsLearnedThisIterConverted = new Dictionary <SemgrexPattern, E>();
SemgrexPattern pat = SemgrexPattern.Compile(pattern.ToString(notAllowedClasses), new Env(constVars.env[label].GetVariables()));
depPatternsLearnedThisIterConverted[pat] = pattern;
}
else
{
throw new NotSupportedException();
}
}
//Apply the patterns and extract candidate phrases
int num;
int numThreads = constVars.numThreads;
//If number of sentences is less, do not create so many threads
if (sents.Count < 50)
{
numThreads = 1;
}
if (numThreads == 1)
{
num = sents.Count;
}
else
{
num = sents.Count / (numThreads - 1);
}
IExecutorService executor = Executors.NewFixedThreadPool(constVars.numThreads);
IList <IFuture <Triple <TwoDimensionalCounter <CandidatePhrase, E>, CollectionValuedMap <E, Triple <string, int, int> >, ICollection <CandidatePhrase> > > > list = new List <IFuture <Triple <TwoDimensionalCounter <CandidatePhrase, E>, CollectionValuedMap <E
, Triple <string, int, int> >, ICollection <CandidatePhrase> > > >();
for (int i = 0; i < numThreads; i++)
{
ICallable <Triple <TwoDimensionalCounter <CandidatePhrase, E>, CollectionValuedMap <E, Triple <string, int, int> >, ICollection <CandidatePhrase> > > task = null;
if (pattern.type.Equals(PatternFactory.PatternType.Surface))
{
//Redwood.log(Redwood.DBG, "Applying pats: assigning sentences " + i*num + " to " +Math.min(sentids.size(), (i + 1) * num) + " to thread " + (i+1));
task = new ApplyPatterns(sents, num == sents.Count ? sentids : sentids.SubList(i * num, Math.Min(sentids.Count, (i + 1) * num)), surfacePatternsLearnedThisIterConverted, label, constVars.removeStopWordsFromSelectedPhrases, constVars.removePhrasesWithStopWords
, constVars);
}
else
{
task = new ApplyDepPatterns(sents, num == sents.Count ? sentids : sentids.SubList(i * num, Math.Min(sentids.Count, (i + 1) * num)), depPatternsLearnedThisIterConverted, label, constVars.removeStopWordsFromSelectedPhrases, constVars.removePhrasesWithStopWords
, constVars);
}
IFuture <Triple <TwoDimensionalCounter <CandidatePhrase, E>, CollectionValuedMap <E, Triple <string, int, int> >, ICollection <CandidatePhrase> > > submit = executor.Submit(task);
list.Add(submit);
}
// Now retrieve the result
foreach (IFuture <Triple <TwoDimensionalCounter <CandidatePhrase, E>, CollectionValuedMap <E, Triple <string, int, int> >, ICollection <CandidatePhrase> > > future in list)
{
try
{
Triple <TwoDimensionalCounter <CandidatePhrase, E>, CollectionValuedMap <E, Triple <string, int, int> >, ICollection <CandidatePhrase> > result = future.Get();
Redwood.Log(ConstantsAndVariables.extremedebug, "Pattern " + pattern + " extracted phrases " + result.First());
wordsandLemmaPatExtracted.AddAll(result.First());
matchedTokensByPat.AddAll(result.Second());
Sharpen.Collections.AddAll(alreadyLabeledWords, result.Third());
}
catch (Exception e)
{
executor.ShutdownNow();
throw new Exception(e);
}
}
executor.Shutdown();
}
