known_item_search_classifier 0.1.0 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: df03a8dc0661439a41c7d3366a49573f463a176f
-  data.tar.gz: 58713a770bba4a173adc4e88bb507f256b4eafc5
+  metadata.gz: 9723eb411c5aac044ace45f88b1529d56b5cbfb2
+  data.tar.gz: 34757a78f9268eb5b88d1fdb0bea1a9c1569e32d
 SHA512:
-  metadata.gz: 470565b3b2932df41a0d02b99048746b5f5e97de476b273a9648d473d63d7e73e44d038ff2b35ffd1f0bbd413960d35df5ca1ad8e8b8902038aedb57d64225f8
-  data.tar.gz: 7e81f8262925a653ab33c12844947f9df40b5510cf3e43d801d6a31cab7f3bf3a3de58dc9693ecbd71283ff309dc9bdb6490d1a6be25fe436726ffc09fea28e0
+  metadata.gz: 961dc70c84bfcb5c2ddfaa866c489c149d84e09ce0bdbb72dd170c7ebdbf0c800cff8dd22d45ab486a84c48a1cfdde91104f5d1cff7828875885d805db058848
+  data.tar.gz: 7b4ca3cd5cb83a9380da7509a2758eb3a6b7e1f414ee44e2a58f809ebfab68cace4ce8d6c3c94b2c0bb6676285633d91f9f9517a489168b2847e854d3b4e2160

data/lib/known_item_search_classifier/classifier.rb

@@ -1,8 +1,10 @@
 # Classifies search strings as either known-item searches or unknown-item searches
+require 'csv'
 require 'gaussian_naive_bayes'
 
 module KnownItemSearchClassifier
     class Classifier
+        attr_accessor :custom_training_set
         def initialize
             set = DefaultTrainingSet.new
             @default_training_set = GaussianNaiveBayes::Classifier.new set.categories_summaries, set.categories_probabilities
@@ -11,7 +13,7 @@ module KnownItemSearchClassifier
             return classify query_string
         end
         def train training_set
-            if defined? @custom_training_set
+            unless defined? @custom_training_set
                 @custom_training_set = GaussianNaiveBayes::Learner.new
             end
             training_set.each do |query|
@@ -19,10 +21,10 @@ module KnownItemSearchClassifier
             end
         end
         def train_from_csv filename
-            if defined? @custom_training_set
+            unless defined? @custom_training_set
                 @custom_training_set = GaussianNaiveBayes::Learner.new
             end
-            csv = CSV.read(filename)
+            csv = ::CSV.read(filename)
             csv.each do |line|
                 submit_vector line
             end

data/lib/known_item_search_classifier/default_training_set.rb

@@ -2,20 +2,22 @@ module KnownItemSearchClassifier
     class DefaultTrainingSet
         attr_reader :categories_probabilities, :categories_summaries
         def initialize
-            @categories_probabilities={:unknown=>0.78, :known=>0.22}
+            @categories_probabilities={:unknown=>0.835, :known=>0.165}
             @categories_summaries= {
                 :unknown=>{
-                    0=>{:mean=>0.2564102564102564, :standard_deviation=>0.4394771815921655},
-                    1=>{:mean=>0.03418803418803419, :standard_deviation=>0.11344969312798027},
-                    2=>{:mean=>0.002564102564102564, :standard_deviation=>0.0226455406828919},
-                    3=>{:mean=>0.12991452991452992, :standard_deviation=>0.26648206508636013},
-                    4=>{:mean=>2.7948717948717947, :standard_deviation=>2.053561836691609}},
+                    0=>{:mean=>0.32335329341317365, :standard_deviation=>0.4691630728112455},
+                    1=>{:mean=>0.01867693185058454, :standard_deviation=>0.0856521002382124},
+                    2=>{:mean=>0.0024950099800399197, :standard_deviation=>0.02318575984424029},
+                    3=>{:mean=>0.18252067293983462, :standard_deviation=>0.32649287803592736},
+                    4=>{:mean=>2.2634730538922154, :standard_deviation=>1.3497147972472143},
+                    5=>{:mean=>0.20958083832335328, :standard_deviation=>1.2933208182456999}},
                 :known=>{
-                    0=>{:mean=>0.5454545454545454, :standard_deviation=>0.5096471914376255},
-                    1=>{:mean=>0.051659451659451655, :standard_deviation=>0.07957404805575267},
-                    2=>{:mean=>0.021248196248196245, :standard_deviation=>0.04412470821426937},
-                    3=>{:mean=>0.22550505050505054, :standard_deviation=>0.2520704609787127},
-                    4=>{:mean=>7.590909090909091, :standard_deviation=>5.770690236086651}}}
+                    0=>{:mean=>0.3333333333333333, :standard_deviation=>0.478713553878169},
+                    1=>{:mean=>0.034283854046699896, :standard_deviation=>0.07844034834013752},
+                    2=>{:mean=>0.06397250092902267, :standard_deviation=>0.10673099909054994},
+                    3=>{:mean=>0.06715805055726004, :standard_deviation=>0.1488979015655406},
+                    4=>{:mean=>4.696969696969697, :standard_deviation=>4.9591131294116515},
+                    5=>{:mean=>3.9393939393939394, :standard_deviation=>5.606577576491037}}}
         end
     end
 end

data/lib/known_item_search_classifier/feature_extractor.rb

@@ -12,13 +12,14 @@ module KnownItemSearchClassifier
             @punctuation_ratio = punctuation_ratio
             @determiner_ratio = determiner_ratio
             @proper_noun_ratio = proper_noun_ratio
+            @numeric_count = numeric_count
 
             #@num_keywords = count_keywords
             #@refers_to_an_item_that_is_known = check_against_known_titles
             
         end
         def feature_array
-            return [@mixed_case, @punctuation_ratio, @determiner_ratio, @proper_noun_ratio, @num_words]
+            return [@mixed_case, @punctuation_ratio, @determiner_ratio, @proper_noun_ratio, @num_words, @numeric_count]
         end
         private
         def is_mixed_case?
@@ -35,6 +36,9 @@ module KnownItemSearchClassifier
             num_det = @tagged.scan(/\/DET/).size.to_f
             return num_det / @num_words
         end
+        def numeric_count
+            return @string.scan(/[0-9]/).length
+        end
         def proper_noun_ratio
             num_prop_noun = @tagged.scan(/\/NNP/).size.to_f
             return num_prop_noun / @num_words
@@ -55,6 +59,7 @@ module KnownItemSearchClassifier
         end
         def check_against_known_titles
            known_titles = [
+               'fountainhead',
                'salt sugar fat',
            ]
            if known_titles.include? @query_string.downcase

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: known_item_search_classifier
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.0
 platform: ruby
 authors:
 - Jane Sandberg
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-11-13 00:00:00.000000000 Z
+date: 2017-03-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: engtagger