Name Entity Extraction from Text

lib/alchemy_api.rb

@@ -9,6 +9,7 @@
 require 'alchemy_api/language_detection'
 require 'alchemy_api/term_extraction'
 require 'alchemy_api/text_extraction'
+require 'alchemy_api/named_entity_extraction'
 
 module AlchemyApi
   @api_key = nil

lib/alchemy_api/concept_tagging.rb

@@ -1,7 +1,6 @@
 # TODO: add support for linkedData return fields
 module AlchemyApi
   ConceptTaggingResult = Struct.new(:concepts, :language, :url, :source_text)
-  Concept = Struct.new(:text, :relevance)
 
   class ConceptTagging < Base
     # http://www.alchemyapi.com/api/concept/textc.html
@@ -51,10 +50,8 @@ class ConceptTagging < Base
 
     def self.get_concepts_handler(response)
       json = get_json(response)
-      concepts = json['concepts'].map do |c|
-        Concept.new(c['text'], c['relevance'].to_f)
-      end
-      ConceptTaggingResult.new(concepts, json['language'], json['url'], json['text'])
+
+      ConceptTaggingResult.new(json['concepts'], json['language'], json['url'], json['text'])
     end
   end
 end

lib/alchemy_api/named_entity_extraction.rb

@@ -0,0 +1,25 @@
+module AlchemyApi
+  NamedEntitiesResult = Struct.new(:entities, :language, :url, :source_text)
+
+  class NamedEntityExtraction < Base
+
+    post(:get_named_entities_from_text) do |text, *args|
+      options = args.first || {}
+      uri "#{AlchemyApi.base_text_uri}/TextGetRankedNamedEntities"
+      params  :text => text,
+              :url => options[:url],
+              :maxRetrieve => options[:max_retrieve] || 10,
+              :showSourceText => options[:show_source_text] ? 1 : 0
+      handler do |response|
+        AlchemyApi::NamedEntityExtraction.get_ranked_entities_handler(response)
+      end
+    end
+
+    def self.get_ranked_entities_handler(response)
+      json = get_json(response)
+
+      NamedEntitiesResult.new(json['entities'], json['language'],
+                               json['url'], json['text'])
+    end
+  end
+end

lib/alchemy_api/term_extraction.rb

@@ -1,6 +1,5 @@
 module AlchemyApi
   TermExtractionResult = Struct.new(:keywords, :language, :url, :source_text)
-  Keyword = Struct.new(:text, :relevance)
 
   class TermExtraction < Base
     post(:get_ranked_keywords_from_html) do |html, *args|
@@ -46,10 +45,8 @@ class TermExtraction < Base
 
     def self.get_ranked_keywords_handler(response)
       json = get_json(response)
-      keywords = json['keywords'].map do |kw|
-        Keyword.new(kw['text'], kw['relevance'].to_f)
-      end
-      TermExtractionResult.new(keywords, json['language'],
+
+      TermExtractionResult.new(json['keywords'], json['language'],
                                json['url'], json['text'])
     end
   end

spec/alchemy_api/named_entity_extraction_spec.rb

@@ -0,0 +1,24 @@
+require File.dirname(__FILE__) + "/../spec_helper"
+
+describe AlchemyApi::NamedEntityExtraction do
+
+  typhoeus_spec_cache('spec/cache/named_entity_extraction/get_named_entity_text') do |hydra|
+    describe "#get_named_entities_from_text" do
+      before(:each) do
+        @url = "http://test.com"
+        text = fixture_for('article.txt')
+        @result = AlchemyApi::NamedEntityExtraction.
+          get_named_entities_from_text(text,
+                                        :url => @url,
+                                        :max_retrieve => 5,
+                                        :show_source_text => true)
+      end
+
+      it "should return at least one entity" do
+        @results.entites.should_not be_empty
+      end
+    end
+  end
+
+end
+