bayon 0.1.0 → 0.1.1

data/README

@@ -31,6 +31,15 @@ gem install bayon
     result.each do |labels|
       puts labels.join(', ')
     end
+    
+    docs.output_similairty_point = true
+    result = docs.do_clustering
+    
+    result.each do |label_points|
+      puts label_points.map {|label, point|
+        "#{label}(#{point})"
+      }.join(', ')
+    end
 
 == Project Page
 

data/ext/bayonext.cpp

@@ -85,6 +85,8 @@ VALUE CBayonDocument::rb_cBayonDocument = Qnil;
 class CBayonAnalyzer {
   bayon::Analyzer* analyzer_;
 
+  bool output_similairty_point_;
+
   static void free(CBayonAnalyzer *p) {
     if (p->analyzer_) {
       delete p->analyzer_;
@@ -107,6 +109,7 @@ class CBayonAnalyzer {
 
     Data_Get_Struct(self, CBayonAnalyzer, p);
     p->analyzer_ = new bayon::Analyzer;
+    p->output_similairty_point_ = false;
 
     return Qnil;
   }
@@ -144,6 +147,24 @@ class CBayonAnalyzer {
     return Qnil;
   }
 
+  static VALUE set_output_similairty_point(VALUE self, VALUE v_output) {
+    CBayonAnalyzer *p;
+    bool output = false;
+
+    if (TYPE(v_output) == T_TRUE) {
+      output = true;
+    } else if (TYPE(v_output) == T_FALSE) {
+      output = false;
+    } else {
+      rb_raise(rb_eTypeError, "wrong argument type %s (expected boolean value)");
+    }
+
+    Data_Get_Struct(self, CBayonAnalyzer, p);
+    p->output_similairty_point_ = output;
+
+    return Qnil;
+  }
+
   static VALUE do_clustering(VALUE self, VALUE v_method) {
     CBayonAnalyzer *p;
 
@@ -156,18 +177,27 @@ class CBayonAnalyzer {
   }
 
   static VALUE get_next_result(VALUE self) {
+    typedef std::vector< std::pair<bayon::Document *, double> > documents;
     CBayonAnalyzer *p;
     bayon::Cluster cluster;
 
     Data_Get_Struct(self, CBayonAnalyzer, p);
 
     if(p->analyzer_->get_next_result(cluster)) {
-      const std::vector<bayon::Document *> documents = cluster.documents();
-      VALUE docids = rb_ary_new2(documents.size());
-
-      for(std::vector<bayon::Document *>::const_iterator i = documents.begin(); i != documents.end(); i++) {
-        bayon::Document* doc = *i;
-        rb_ary_push(docids, LONG2NUM(doc->id()));
+      std::vector< std::pair<bayon::Document *, double> > pairs;
+      cluster.sorted_documents(pairs);
+      VALUE docids = rb_ary_new2(pairs.size());
+
+      for(documents::const_iterator i = pairs.begin(); i != pairs.end(); i++) {
+        bayon::Document* doc = i->first;
+        double point = i->second;
+
+        if (p->output_similairty_point_) {
+          VALUE docid_points = rb_ary_new3(2, LONG2NUM(doc->id()), rb_float_new(point));
+          rb_ary_push(docids, docid_points);
+        } else {
+          rb_ary_push(docids, LONG2NUM(doc->id()));
+        }
       }
 
       return docids;
@@ -185,6 +215,7 @@ public:
     rb_define_method(rb_cBayonAnalyzer, "add_document", __F(&add_document), 1);
     rb_define_method(rb_cBayonAnalyzer, "set_cluster_size_limit", __F(&set_cluster_size_limit), 1);
     rb_define_method(rb_cBayonAnalyzer, "set_eval_limit", __F(&set_eval_limit), 1);
+    rb_define_method(rb_cBayonAnalyzer, "set_output_similairty_point", __F(&set_output_similairty_point), 1);
     rb_define_method(rb_cBayonAnalyzer, "do_clustering", __F(&do_clustering), 1);
     rb_define_method(rb_cBayonAnalyzer, "get_next_result", __F(&get_next_result), 0);
 

data/lib/bayon.rb

@@ -6,6 +6,7 @@ module Bayon
       @documents = []
       @cluster_size_limit = nil
       @eval_limit = nil
+      @output_similairty_point = nil
     end
 
     def cluster_size_limit=(limit)
@@ -24,6 +25,14 @@ module Bayon
       @eval_limit = limit
     end
 
+    def output_similairty_point=(output)
+      unless output.instance_of?(TrueClass) or output.instance_of?(FalseClass)
+        raise TypeError, "wrong argument type #{limit.class} (expected boolean value)"
+      end
+
+      @output_similairty_point = output
+    end
+    
     def add_document(label, features)
       unless features.kind_of?(Hash)
         raise TypeError, "wrong argument type #{limit.class} (expected Hash)"
@@ -40,6 +49,7 @@ module Bayon
       analyzer = Analyzer.new
       analyzer.set_cluster_size_limit(@cluster_size_limit) if @cluster_size_limit
       analyzer.set_eval_limit(@eval_limit) if @eval_limit
+      analyzer.set_output_similairty_point(@output_similairty_point) if @output_similairty_point
 
       feature_set = []
 
@@ -59,7 +69,11 @@ module Bayon
       result = []
 
       while (cluster = analyzer.get_next_result)
-        result << cluster.map {|doc_id| @documents[doc_id][0] }
+        if @output_similairty_point
+          result << cluster.map {|doc_id, point| [@documents[doc_id][0], point] }
+        else
+          result << cluster.map {|doc_id| @documents[doc_id][0] }
+        end
       end
 
       return result

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: bayon
 version: !ruby/object:Gem::Version 
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors: 
 - winebarrel
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-06-14 00:00:00 +09:00
+date: 2009-06-24 00:00:00 +09:00
 default_executable: 
 dependencies: []