Updated config file handling for pyp-topics and pipeline.

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@141 ec762483-ff6d-05da-a07a-a48fb63a330f

7 files changed, 102 insertions, 37 deletions

diff --git a/gi/pyp-topics/src/contexts_corpus.cc b/gi/pyp-topics/src/contexts_corpus.cc
index f01d352a..280b2976 100644
--- a/gi/pyp-topics/src/contexts_corpus.cc
+++ b/gi/pyp-topics/src/contexts_corpus.cc
@@ -23,7 +23,7 @@ void read_callback(const ContextsLexer::PhraseContextsType& new_contexts, void*
 
   ContextsCorpus* corpus_ptr = extra_pair->get<0>();
   BackoffGenerator* backoff_gen = extra_pair->get<1>();
-  map<string,int>* counts = extra_pair->get<2>();
+  //map<string,int>* counts = extra_pair->get<2>();
 
   Document* doc(new Document());
 
@@ -33,11 +33,11 @@ void read_callback(const ContextsLexer::PhraseContextsType& new_contexts, void*
     string context_str = corpus_ptr->m_dict.toString(new_contexts.contexts[i]);
 
     // filter out singleton contexts
-    if (!counts->empty()) {
-      map<string,int>::const_iterator find_it = counts->find(context_str);
-      if (find_it == counts->end() || find_it->second < 2)
-        continue;
-    }
+    //if (!counts->empty()) {
+    //  map<string,int>::const_iterator find_it = counts->find(context_str);
+    //  if (find_it == counts->end() || find_it->second < 2)
+    //    continue;
+    //}
 
     WordID id = corpus_ptr->m_dict.Convert(context_str);
     if (cache_word_count != corpus_ptr->m_dict.max()) {
@@ -85,10 +85,10 @@ void read_callback(const ContextsLexer::PhraseContextsType& new_contexts, void*
   }
   //cout << endl;
 
-  if (!doc->empty()) {
+  //if (!doc->empty()) {
     corpus_ptr->m_documents.push_back(doc);
     corpus_ptr->m_keys.push_back(new_contexts.phrase);
-  }
+  //}
 }
 
 void filter_callback(const ContextsLexer::PhraseContextsType& new_contexts, void* extra) {
@@ -108,10 +108,12 @@ void filter_callback(const ContextsLexer::PhraseContextsType& new_contexts, void
 
 unsigned ContextsCorpus::read_contexts(const string &filename, 
                                        BackoffGenerator* backoff_gen_ptr,
-                                       bool filter_singeltons) {
+                                       bool /*filter_singeltons*/) {
   map<string,int> counts;
-  if (filter_singeltons) {
-    cerr << "--- Filtering singleton contexts ---" << endl;
+  //if (filter_singeltons) 
+  {
+  //  cerr << "--- Filtering singleton contexts ---" << endl;
+
     igzstream in(filename.c_str());
     ContextsLexer::ReadContexts(&in, filter_callback, &counts);
   }
@@ -128,7 +130,15 @@ unsigned ContextsCorpus::read_contexts(const string &filename,
   cerr << "Read backoff with order " << m_backoff->order() << "\n";
   for (int o=0; o<m_backoff->order(); o++)
     cerr << "  Terms at " << o << " = " << m_backoff->terms_at_level(o) << endl;
-  cerr << endl;
+  //cerr << endl;
+
+  int i=0; double av_freq=0;
+  for (map<string,int>::const_iterator it=counts.begin(); it != counts.end(); ++it, ++i) {
+    WordID id = m_dict.Convert(it->first);
+    m_context_counts[id] = it->second;
+    av_freq += it->second;
+  }
+  cerr << "  Average term frequency = " << av_freq / (double) i << endl;
 
   return m_documents.size();
 }
diff --git a/gi/pyp-topics/src/contexts_corpus.hh b/gi/pyp-topics/src/contexts_corpus.hh
index 891e3a6b..66b71783 100644
--- a/gi/pyp-topics/src/contexts_corpus.hh
+++ b/gi/pyp-topics/src/contexts_corpus.hh
@@ -4,6 +4,7 @@
 #include <vector>
 #include <string>
 #include <map>
+#include <tr1/unordered_map>
 
 #include <boost/ptr_container/ptr_vector.hpp>
 
@@ -66,6 +67,11 @@ public:
       return res;
     }
 
+    virtual int context_count(const WordID& id) const {
+      return m_context_counts.find(id)->second;
+    }
+
+
     const std::string& key(const int& i) const {
       return m_keys.at(i);
     }
@@ -74,6 +80,7 @@ private:
     TermBackoffPtr m_backoff;
     Dict m_dict;
     std::vector<std::string> m_keys;
+    std::tr1::unordered_map<int,int> m_context_counts;
 };
 
 #endif // _CONTEXTS_CORPUS_HH
diff --git a/gi/pyp-topics/src/corpus.hh b/gi/pyp-topics/src/corpus.hh
index c2f37130..24981946 100644
--- a/gi/pyp-topics/src/corpus.hh
+++ b/gi/pyp-topics/src/corpus.hh
@@ -4,6 +4,7 @@
 #include <vector>
 #include <string>
 #include <map>
+#include <limits>
 
 #include <boost/shared_ptr.hpp>
 #include <boost/ptr_container/ptr_vector.hpp>
@@ -35,6 +36,10 @@ public:
     int num_terms() const { return m_num_terms; }
     int num_types() const { return m_num_types; }
 
+    virtual int context_count(const int&) const {
+      return std::numeric_limits<int>::max();
+    }
+
 protected:
     int m_num_terms, m_num_types;
     boost::ptr_vector<Document> m_documents; 
diff --git a/gi/pyp-topics/src/pyp-topics.cc b/gi/pyp-topics/src/pyp-topics.cc
index 4fb75caa..0ac1b709 100644
--- a/gi/pyp-topics/src/pyp-topics.cc
+++ b/gi/pyp-topics/src/pyp-topics.cc
@@ -29,7 +29,9 @@ struct Timer {
   timespec start_t;
 };
 
-void PYPTopics::sample(const Corpus& corpus, int samples) {
+void PYPTopics::sample_corpus(const Corpus& corpus, int samples, 
+                              int freq_cutoff_start, int freq_cutoff_end, 
+                              int freq_cutoff_interval) {
   Timer timer;
 
   if (!m_backoff.get()) {
@@ -37,7 +39,7 @@ void PYPTopics::sample(const Corpus& corpus, int samples) {
     m_word_pyps.push_back(PYPs());
   }
 
-  std::cerr << " Training with " << m_word_pyps.size()-1 << " backoff level"
+  std::cerr << "\n Training with " << m_word_pyps.size()-1 << " backoff level"
     << (m_word_pyps.size()==2 ? ":" : "s:") << std::endl;
 
   for (int i=0; i<(int)m_word_pyps.size(); ++i)
@@ -53,6 +55,9 @@ void PYPTopics::sample(const Corpus& corpus, int samples) {
   std::cerr << " Documents: " << corpus.num_documents() << " Terms: "
     << corpus.num_types() << std::endl;
 
+  int frequency_cutoff = freq_cutoff_start;
+  std::cerr << " Context frequency cutoff set to " << frequency_cutoff << std::endl; 
+
   timer.Reset();
   // Initialisation pass
   int document_id=0, topic_counter=0;
@@ -68,19 +73,22 @@ void PYPTopics::sample(const Corpus& corpus, int samples) {
 
       // sample a new_topic
       //int new_topic = (topic_counter % m_num_topics);
-      int new_topic = (document_id % m_num_topics);
+      int freq = corpus.context_count(term);
+      int new_topic = (freq > frequency_cutoff ? (document_id % m_num_topics) : -1);
 
       // add the new topic to the PYPs
       m_corpus_topics[document_id][term_index] = new_topic;
-      increment(term, new_topic);
+      if (freq > frequency_cutoff) {
+        increment(term, new_topic);
 
-      if (m_use_topic_pyp) {
-        F p0 = m_topic_pyp.prob(new_topic, m_topic_p0);
-        int table_delta = m_document_pyps[document_id].increment(new_topic, p0);
-        if (table_delta)
-          m_topic_pyp.increment(new_topic, m_topic_p0);
+        if (m_use_topic_pyp) {
+          F p0 = m_topic_pyp.prob(new_topic, m_topic_p0);
+          int table_delta = m_document_pyps[document_id].increment(new_topic, p0);
+          if (table_delta)
+            m_topic_pyp.increment(new_topic, m_topic_p0);
+        }
+        else m_document_pyps[document_id].increment(new_topic, m_topic_p0);
       }
-      else m_document_pyps[document_id].increment(new_topic, m_topic_p0);
     }
   }
   std::cerr << "  Initialized in " << timer.Elapsed() << " seconds\n";
@@ -91,6 +99,13 @@ void PYPTopics::sample(const Corpus& corpus, int samples) {
 
   // Sampling phase
   for (int curr_sample=0; curr_sample < samples; ++curr_sample) {
+    if (freq_cutoff_interval > 0 && curr_sample != 1 
+        && curr_sample % freq_cutoff_interval == 1 
+        && frequency_cutoff > freq_cutoff_end) {
+      frequency_cutoff--;
+      std::cerr << "\n Context frequency cutoff set to " << frequency_cutoff << std::endl; 
+    }
+
     std::cerr << "\n  -- Sample " << curr_sample << " "; std::cerr.flush();
 
     // Randomize the corpus indexing array
@@ -115,14 +130,20 @@ void PYPTopics::sample(const Corpus& corpus, int samples) {
       for (Document::const_iterator docIt=corpus.at(document_id).begin();
            docIt != docEnd; ++docIt, ++term_index) {
         Term term = *docIt;
+        int freq = corpus.context_count(term);
+        if (freq < frequency_cutoff) 
+          continue;
 
         // remove the prevous topic from the PYPs
         int current_topic = m_corpus_topics[document_id][term_index];
-        decrement(term, current_topic);
+        // a negative label mean that term hasn't been sampled yet
+        if (current_topic >= 0) {
+          decrement(term, current_topic);
 
-        int table_delta = m_document_pyps[document_id].decrement(current_topic);
-        if (m_use_topic_pyp && table_delta < 0)
-          m_topic_pyp.decrement(current_topic);
+          int table_delta = m_document_pyps[document_id].decrement(current_topic);
+          if (m_use_topic_pyp && table_delta < 0) 
+            m_topic_pyp.decrement(current_topic);
+        }
 
         // sample a new_topic
         int new_topic = sample(document_id, term);
@@ -182,9 +203,9 @@ void PYPTopics::sample(const Corpus& corpus, int samples) {
       std::cerr.precision(2);
       for (PYPs::iterator pypIt=m_word_pyps.front().begin();
            pypIt != m_word_pyps.front().end(); ++pypIt, ++k) {
-        std::cerr << "<" << k << ":" << pypIt->num_customers() << ","
-          << pypIt->num_types() << "," << m_topic_pyp.prob(k, m_topic_p0) << "> ";
         if (k % 5 == 0) std::cerr << std::endl << '\t';
+        std::cerr << "<" << k << ":" << pypIt->num_customers() << "," 
+          << pypIt->num_types() << "," << m_topic_pyp.prob(k, m_topic_p0) << "> ";
       }
       std::cerr.precision(4);
       std::cerr << std::endl;
diff --git a/gi/pyp-topics/src/pyp-topics.hh b/gi/pyp-topics/src/pyp-topics.hh
index c35645aa..d4d87440 100644
--- a/gi/pyp-topics/src/pyp-topics.hh
+++ b/gi/pyp-topics/src/pyp-topics.hh
@@ -19,7 +19,10 @@ public:
     : m_num_topics(num_topics), m_word_pyps(1), 
     m_topic_pyp(0.5,1.0), m_use_topic_pyp(use_topic_pyp) {}
 
-  void sample(const Corpus& corpus, int samples);
+  void sample_corpus(const Corpus& corpus, int samples,
+                     int freq_cutoff_start=0, int freq_cutoff_end=0, 
+                     int freq_cutoff_interval=0);
+
   int sample(const DocumentId& doc, const Term& term);
   int max(const DocumentId& doc, const Term& term) const;
   int max(const DocumentId& doc) const;
diff --git a/gi/pyp-topics/src/train-contexts.cc b/gi/pyp-topics/src/train-contexts.cc
index 7e2100f8..481f8926 100644
--- a/gi/pyp-topics/src/train-contexts.cc
+++ b/gi/pyp-topics/src/train-contexts.cc
@@ -34,8 +34,13 @@ int main(int argc, char **argv)
 
   // Command line processing
   {
-    options_description cmdline_options("Allowed options");
-    cmdline_options.add_options()
+    options_description cmdline_specific("Command line specific options");
+    cmdline_specific.add_options()
+      ("help,h", "print help message")
+      ("config,c", value<string>(), "config file specifying additional command line options")
+      ;
+    options_description config_options("Allowed options");
+    config_options.add_options()
       ("help,h", "print help message")
       ("data,d", value<string>(), "file containing the documents and context terms")
       ("topics,t", value<int>()->default_value(50), "number of topics")
@@ -44,14 +49,25 @@ int main(int argc, char **argv)
       ("topic-words-out,w", value<string>(), "file to write the topic word distribution to")
       ("samples,s", value<int>()->default_value(10), "number of sampling passes through the data")
       ("backoff-type", value<string>(), "backoff type: none|simple")
-      ("filter-singleton-contexts", "filter singleton contexts")
+//      ("filter-singleton-contexts", "filter singleton contexts")
       ("hierarchical-topics", "Use a backoff hierarchical PYP as the P0 for the document topics distribution.")
+      ("freq-cutoff-start", value<int>()->default_value(0), "initial frequency cutoff.")
+      ("freq-cutoff-end", value<int>()->default_value(0), "final frequency cutoff.")
+      ("freq-cutoff-interval", value<int>()->default_value(0), "number of iterations between frequency decrement.")
       ;
-    store(parse_command_line(argc, argv, cmdline_options), vm); 
+
+    cmdline_specific.add(config_options);
+
+    store(parse_command_line(argc, argv, cmdline_specific), vm); 
     notify(vm);
 
+    if (vm.count("config") > 0) {
+      ifstream config(vm["config"].as<string>().c_str());
+      store(parse_config_file(config, config_options), vm); 
+    }
+
     if (vm.count("help")) { 
-      cout << cmdline_options << "\n"; 
+      cout << cmdline_specific << "\n"; 
       return 1; 
     }
   }
@@ -83,14 +99,17 @@ int main(int argc, char **argv)
   }
 
   ContextsCorpus contexts_corpus;
-  contexts_corpus.read_contexts(vm["data"].as<string>(), backoff_gen, vm.count("filter-singleton-contexts"));
+  contexts_corpus.read_contexts(vm["data"].as<string>(), backoff_gen, /*vm.count("filter-singleton-contexts")*/ false);
   model.set_backoff(contexts_corpus.backoff_index());
 
   if (backoff_gen) 
     delete backoff_gen;
 
   // train the sampler
-  model.sample(contexts_corpus, vm["samples"].as<int>());
+  model.sample_corpus(contexts_corpus, vm["samples"].as<int>(),
+                      vm["freq-cutoff-start"].as<int>(),
+                      vm["freq-cutoff-end"].as<int>(),
+               vm["freq-cutoff-interval"].as<int>());
 
   if (vm.count("document-topics-out")) {
     ogzstream documents_out(vm["document-topics-out"].as<string>().c_str());
diff --git a/gi/pyp-topics/src/train.cc b/gi/pyp-topics/src/train.cc
index f7b01af0..c94010f2 100644
--- a/gi/pyp-topics/src/train.cc
+++ b/gi/pyp-topics/src/train.cc
@@ -83,7 +83,7 @@ int main(int argc, char **argv)
     model.set_backoff(vm["backoff-paths"].as<string>());
 
   // train the sampler
-  model.sample(corpus, vm["samples"].as<int>());
+  model.sample_corpus(corpus, vm["samples"].as<int>());
 
   if (vm.count("document-topics-out")) {
     ogzstream documents_out(vm["document-topics-out"].as<string>().c_str());