open_nlp 0.0.3-java → 0.0.4-java

data/README.md

@@ -7,6 +7,7 @@ A JRuby wrapper for the Apache OpenNLP tools library, that allows you execute co
  * named entity extraction
  * chunks detection
  * parsing
+ * document categorization
 
 ## Installation
 
@@ -66,7 +67,7 @@ Then you can create instances of open_nlp classes and use it for your nlp tasks
 ### Parsing
 
     # parser also needs tokenizer model because it uses tokenizer inside parse task
-    parse_model = OpenNlp::Model::Parser.new(File.join(FIXTURES_DIR, "en-parser-chunking.bin"))
+    parse_model = OpenNlp::Model::Parser.new(File.join("nlp_models/en-parser-chunking.bin"))
     token_model = OpenNlp::Model::Tokenizer.new("nlp_models/en-token.bin")
     parser = OpenNlp::Parser.new(parse_model, token_model)
 
@@ -79,6 +80,12 @@ Then you can create instances of open_nlp classes and use it for your nlp tasks
     # you can get code tree structure of parse result by calling
     parse_info.code_tree
 
+### Categorizing
+
+    doccat_model = OpenNlp::Model::Parser.new(File.join("nlp_models/en-doccat.bin"))
+    categorizer = OpenNlp::Categorizer.new(doccat_model)
+    categorizer.categorize("Quick brown fox jumps very bad.")
+
 ## Contributing
 
 1. Fork it

data/lib/open_nlp/categorizer.rb

@@ -0,0 +1,12 @@
+module OpenNlp
+  class Categorizer < Tool
+    self.java_class = Java::opennlp.tools.doccat.DocumentCategorizerME
+
+    def categorize(str)
+      raise ArgumentError, "str must be a String" unless str.is_a?(String)
+
+      outcomes = @j_instance.categorize(str)
+      @j_instance.getBestCategory(outcomes)
+    end
+  end
+end

data/lib/open_nlp/model/categorizer.rb

@@ -0,0 +1,3 @@
+class OpenNlp::Model::Categorizer < OpenNlp::Model
+  self.java_class = Java::opennlp.tools.doccat.DoccatModel
+end

data/lib/open_nlp/version.rb

@@ -1,3 +1,3 @@
 module OpenNlp
-  VERSION = '0.0.3'
+  VERSION = '0.0.4'
 end

data/lib/open_nlp.rb

@@ -9,6 +9,7 @@ require 'open_nlp/java_class'
 require 'open_nlp/model'
 
 require 'open_nlp/model/chunker'
+require 'open_nlp/model/categorizer'
 require 'open_nlp/model/detokenizer'
 require 'open_nlp/model/named_entity_detector'
 require 'open_nlp/model/pos_tagger'
@@ -17,6 +18,7 @@ require 'open_nlp/model/tokenizer'
 require 'open_nlp/model/parser'
 
 require 'open_nlp/tool'
+require 'open_nlp/categorizer'
 require 'open_nlp/named_entity_detector'
 require 'open_nlp/pos_tagger'
 require 'open_nlp/sentence_detector'

data/spec/categorizer_spec.rb

@@ -0,0 +1,36 @@
+require 'spec_helper'
+
+describe OpenNlp::Categorizer do
+  subject { OpenNlp::Categorizer }
+  let(:model){ OpenNlp::Model::Categorizer.new(File.join(FIXTURES_DIR, "en-doccat.bin")) }
+
+  describe "initialization" do
+    it "should initialize with a valid model" do
+      categorizer = subject.new(model)
+      categorizer.should be_a(subject)
+      categorizer.j_instance.should be_a(subject.java_class)
+    end
+
+    it "should raise an ArgumentError without a valid model" do
+      lambda { subject.new(nil) }.should raise_error(ArgumentError)
+    end
+  end
+
+  describe "categorizing a string" do
+    let(:categorizer) { subject.new(model) }
+
+    it "should categorize a provided document to positive" do
+      category = categorizer.categorize("The fox is a good worker.")
+      category.should == "Positive"
+    end
+
+    it "should categorize a provided document to negative" do
+      category = categorizer.categorize("Quick brown fox jumps very bad.")
+      category.should == "Negative"
+    end
+
+    it "should raise an ArgumentError for a non-string" do
+      lambda { categorizer.categorize(nil) }.should raise_error(ArgumentError)
+    end
+  end
+end

metadata

@@ -2,7 +2,7 @@
 name: open_nlp
 version: !ruby/object:Gem::Version
   prerelease:
-  version: 0.0.3
+  version: 0.0.4
 platform: java
 authors:
 - Hck
@@ -24,9 +24,11 @@ files:
 - README.md
 - Rakefile
 - lib/open_nlp.rb
+- lib/open_nlp/categorizer.rb
 - lib/open_nlp/chunker.rb
 - lib/open_nlp/java_class.rb
 - lib/open_nlp/model.rb
+- lib/open_nlp/model/categorizer.rb
 - lib/open_nlp/model/chunker.rb
 - lib/open_nlp/model/detokenizer.rb
 - lib/open_nlp/model/named_entity_detector.rb
@@ -43,12 +45,13 @@ files:
 - lib/open_nlp/sentence_detector.rb
 - lib/open_nlp/tokenizer.rb
 - lib/open_nlp/tool.rb
-- lib/open_nlp/utils/span.rb
 - lib/open_nlp/version.rb
 - open_nlp.gemspec
+- spec/categorizer_spec.rb
 - spec/chunker_spec.rb
 - spec/fixtures/en-chunker.bin
 - spec/fixtures/en-detokenizer.xml
+- spec/fixtures/en-doccat.bin
 - spec/fixtures/en-ner-time.bin
 - spec/fixtures/en-parser-chunking.bin
 - spec/fixtures/en-pos-maxent.bin
@@ -94,9 +97,11 @@ signing_key:
 specification_version: 3
 summary: A JRuby wrapper for the Apache OpenNLP tools library
 test_files:
+- spec/categorizer_spec.rb
 - spec/chunker_spec.rb
 - spec/fixtures/en-chunker.bin
 - spec/fixtures/en-detokenizer.xml
+- spec/fixtures/en-doccat.bin
 - spec/fixtures/en-ner-time.bin
 - spec/fixtures/en-parser-chunking.bin
 - spec/fixtures/en-pos-maxent.bin

data/lib/open_nlp/utils/span.rb

@@ -1,15 +0,0 @@
-module OpenNlp
-  module Utils
-    class Span
-      include JavaClass
-
-      self.java_class = Java::opennlp.tools.util.Span
-
-      attr_reader :j_instance
-
-      def initialize(start_offset, end_offset)
-        @j_instance = self.class.java_class.new(start_offset, end_offset)
-      end
-    end
-  end
-end