rsemantic 0.1.3 → 0.1.4

data/spec/semantic/transform/lsa_transform_spec.rb

@@ -1,59 +0,0 @@
-require File.dirname(__FILE__) + '/../../spec_helper'
-
-module Semantic
-  describe Transform::LSA do
-
-    tiny_matrix = Linalg::DMatrix.columns([[0.0, 1.0, 0.0],
-                                           [1.0, 0.0, 1.0]])
-
-    u = Linalg::DMatrix.rows([[1,0],
-                              [0,1]])
-
-    vt = Linalg::DMatrix.rows([[1,0,0],
-                               [1,0,0],
-                               [1,0,0]])
-
-    sigma = Linalg::DMatrix.rows([[1,0,0],
-                                  [0,1,0]])
-
-    describe "latent semantic analysis transform" do
-
-      it "should use svd on matrix" do
-        matrix = Linalg::DMatrix.columns([[0.0, 1.0, 0.0],
-                                          [1.0, 0.0, 1.0]])
-
-        matrix.should_receive(:singular_value_decomposition).and_return([u, sigma, vt])
-
-        Linalg::DMatrix.stub!(:columns).and_return(matrix)
-
-        Transform::LSA.transform(matrix)
-      end
-
-      it "should reduce the noise in the sigma matrix" do
-        matrix = Linalg::DMatrix.columns([[0.0, 1.0, 0.0],
-                                          [1.0, 0.0, 1.0]])
-
-        matrix.stub!(:singular_value_decomposition).and_return([u, sigma, vt])
-        Linalg::DMatrix.stub!(:columns).and_return(matrix)
-
-        sigma.should_receive(:[]=).with(0,0,0)
-        sigma.should_receive(:[]=).with(1,1,0)
-
-        Transform::LSA.transform(matrix, 2)
-      end
-
-      it "should prevent reducing dimensions greater than the matrixes own dimensions" do
-        lambda { Transform::LSA.transform tiny_matrix, 100 }.should raise_error(Exception)
-      end
-
-      it "should transform LSA matrix" do
-        transformed_matrix = Transform::LSA.transform tiny_matrix
-
-        #TODO: better way to compare result matrix
-        transformed_matrix.to_s.should == Linalg::DMatrix.columns([[0,0,0],[1,0,1]]).to_s
-      end
-
-    end
-
-  end
-end

data/spec/semantic/transform/tf_idf_transform_spec.rb

@@ -1,35 +0,0 @@
-require File.dirname(__FILE__) + '/../../spec_helper'
-
-module Semantic
-  describe Transform::TFIDF do
-
-    def matrix(matrix)
-      Linalg::DMatrix.rows(matrix)
-    end
-
-    tiny_matrix = Linalg::DMatrix.rows([[0.0, 1.0, 0.0],
-    [1.0, 0.0, 1.0]])
-
-    describe "term frequency / inverse document frequency transform" do
-
-      it "should find the number of times each term occurs" do
-        Transform::TFIDF.should_receive(:number_of_documents_with_term).with(0, matrix([[1]])).and_return(2)
-
-        Transform::TFIDF.transform(matrix([[1]]))
-      end
-
-      it "should ignore counting terms with 0 weighting" do
-        Transform::TFIDF.should_not_receive(:number_of_documents_with_term)
-
-        Transform::TFIDF.transform(matrix([[0,0],[0,0]]))
-      end
-
-      it "should calculate term frequency * inverse document freuency" do
-        transformed_matrix = Transform::TFIDF.transform matrix([[1,1],[0,1]])
-
-        transformed_matrix.to_s.should == Linalg::DMatrix.columns([[0, 0],[0, 0.346574]]).to_s
-      end
-
-    end
-  end
-end

data/spec/semantic/vector_space/builder_spec.rb

@@ -1,44 +0,0 @@
-require File.dirname(__FILE__) + '/../../spec_helper'
-
-module Semantic
-  module VectorSpace
-    describe Builder do
-
-      def mock_parser
-        @parser ||= mock("Parser")
-      end
-
-      def documents
-        ['nipon','ichiban']
-      end
-
-
-      describe "building query vector" do
-
-        it "should build vector from string" do
-          builder = Builder.new
-          builder.should_receive(:build_vector).with("query string")
-
-          builder.build_query_vector(["query","string"])
-        end
-
-        it "should generate a valid vector" do
-          builder = Builder.new
-          builder.build_document_matrix(["query string"])
-          query = builder.build_query_vector(["query","string"])
-
-          query.should == Linalg::DMatrix.columns([[1,1]])
-        end
-
-        it "should generate empty vector when terms are not in document matrix" do
-          builder = Builder.new
-          builder.build_document_matrix(["string"])
-          query = builder.build_query_vector(["not-in-document"])
-
-          query.should == Linalg::DMatrix.columns([[0]])
-        end
-
-      end
-    end
-  end
-end

data/spec/semantic/vector_space/model_spec.rb

@@ -1,22 +0,0 @@
-require File.dirname(__FILE__) + '/../../spec_helper'
-
-module Semantic
-  module VectorSpace
-
-    describe Model do
-
-      it "should output a DMatrix as a pretty string" do
-        model = Model.new(Linalg::DMatrix.columns([[0.11111,0.66666],[0.33333, 0.001]]), {})
-
-        model.to_s.should include("[ +0.11 +0.33 ]\n[ +0.67 +0.00 ]\n")
-      end
-      
-      it "should output keywords for the matrix rows" do
-        model = Model.new(Linalg::DMatrix.columns([[0]]), {'shiva' => 0})
-        
-        model.to_s.should include("shiva [ +0.00 ]")
-      end
-
-    end
-  end
-end

data/spec/spec.opts

@@ -1,2 +0,0 @@
---colour
---diff

data/spec/spec_helper.rb

@@ -1,7 +0,0 @@
-require 'rubygems'
-gem 'rspec'
-require 'spec'
-
-$:.unshift(File.join(File.dirname(__FILE__), %w[.. lib]))
-
-require 'semantic'