From 125a096aaa58202e17f0f53ab0cca86d51c9304c Mon Sep 17 00:00:00 2001
From: Joey Novak <joey.novak@gmail.com>
Date: Thu, 26 Oct 2017 17:21:53 -0700
Subject: [PATCH 1/3] Made _train and _untrain public so that tokens can be
 bulk trained and untrained and move tokenization outside of the library.

---
 BayesSharp/BayesClassifier.cs | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/BayesSharp/BayesClassifier.cs b/BayesSharp/BayesClassifier.cs
index cba4bfe..280220b 100644
--- a/BayesSharp/BayesClassifier.cs
+++ b/BayesSharp/BayesClassifier.cs
@@ -170,7 +170,7 @@ public void Train(TTagType tagId, string input)
         {
             var tokens = _tokenizer.Tokenize(input);
             var tag = GetAndAddIfNotFound(_tags.Items, tagId);
-            _train(tag, tokens);
+            Train(tag, tokens);
             _tags.SystemTag.TrainCount += 1;
             tag.TrainCount += 1;
             _mustRecache = true;
@@ -189,7 +189,7 @@ public void Untrain(TTagType tagId, string input)
             {
                 return;
             }
-            _untrain(tag, tokens);
+            Untrain(tag, tokens);
             _tags.SystemTag.TrainCount += 1;
             tag.TrainCount += 1;
             _mustRecache = true;
@@ -219,7 +219,7 @@ public Dictionary<TTagType, double> Classify(string input)
 
         #region Private Methods
 
-        private void _train(TagData<TTokenType> tag, IEnumerable<TTokenType> tokens)
+        private void Train(TagData<TTokenType> tag, IEnumerable<TTokenType> tokens)
         {
             var tokenCount = 0;
             foreach (var token in tokens)
@@ -234,7 +234,7 @@ private void _train(TagData<TTokenType> tag, IEnumerable<TTokenType> tokens)
             _tags.SystemTag.TokenCount += tokenCount;
         }
 
-        private void _untrain(TagData<TTokenType> tag, IEnumerable<TTokenType> tokens)
+        public void Untrain(TagData<TTokenType> tag, IEnumerable<TTokenType> tokens)
         {
             foreach (var token in tokens)
             {

From f9efc4db6d8bf7d83e4d475241a8b70987238472 Mon Sep 17 00:00:00 2001
From: Joey Novak <joey.novak@gmail.com>
Date: Thu, 26 Oct 2017 17:36:15 -0700
Subject: [PATCH 2/3] Whoops, forgot to make Train() a public method.

---
 BayesSharp/BayesClassifier.cs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/BayesSharp/BayesClassifier.cs b/BayesSharp/BayesClassifier.cs
index 280220b..5a08746 100644
--- a/BayesSharp/BayesClassifier.cs
+++ b/BayesSharp/BayesClassifier.cs
@@ -219,7 +219,7 @@ public Dictionary<TTagType, double> Classify(string input)
 
         #region Private Methods
 
-        private void Train(TagData<TTokenType> tag, IEnumerable<TTokenType> tokens)
+        public void Train(TagData<TTokenType> tag, IEnumerable<TTokenType> tokens)
         {
             var tokenCount = 0;
             foreach (var token in tokens)

From 6838ec4b4cd2f8cdc283dd28f480c2bd2dd7d189 Mon Sep 17 00:00:00 2001
From: Joey Novak <joey.novak@gmail.com>
Date: Thu, 26 Oct 2017 17:40:30 -0700
Subject: [PATCH 3/3] Refactored Classify into a seperate method for tokenized
 vs not tokenized.

---
 BayesSharp/BayesClassifier.cs | 33 +++++++++++++++++++--------------
 1 file changed, 19 insertions(+), 14 deletions(-)

diff --git a/BayesSharp/BayesClassifier.cs b/BayesSharp/BayesClassifier.cs
index 5a08746..aa41f53 100644
--- a/BayesSharp/BayesClassifier.cs
+++ b/BayesSharp/BayesClassifier.cs
@@ -201,23 +201,28 @@ public void Untrain(TTagType tagId, string input)
         /// <param name="input">Input to be classified</param>
         public Dictionary<TTagType, double> Classify(string input)
         {
-            var tokens = _tokenizer.Tokenize(input).ToList();
-            var tags = CreateCacheAnsGetTags();
+           var tokens = _tokenizer.Tokenize(input).ToList();
+           return Classify(tokens);
+        }
 
-            var stats = new Dictionary<TTagType, double>();
+       private Dictionary<TTagType, double> Classify(List<TTokenType> tokens)
+       {
+          var tags = CreateCacheAnsGetTags();
 
-            foreach (var tag in tags.Items)
-            {
-                var probs = GetProbabilities(tag.Value, tokens).ToList();
-                if (probs.Count() != 0)
-                {
-                    stats[tag.Key] = _combiner.Combine(probs);
-                }
-            }
-            return stats.OrderByDescending(s => s.Value).ToDictionary(s => s.Key, pair => pair.Value);
-        }
+          var stats = new Dictionary<TTagType, double>();
+
+          foreach (var tag in tags.Items)
+          {
+             var probs = GetProbabilities(tag.Value, tokens).ToList();
+             if (probs.Count() != 0)
+             {
+                stats[tag.Key] = _combiner.Combine(probs);
+             }
+          }
+          return stats.OrderByDescending(s => s.Value).ToDictionary(s => s.Key, pair => pair.Value);
+       }
 
-        #region Private Methods
+       #region Private Methods
 
         public void Train(TagData<TTokenType> tag, IEnumerable<TTokenType> tokens)
         {