commendo 0.0.7 → 0.0.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: e46e056067c3c4af5ab6315d8145b811286a465c
-  data.tar.gz: 5392a1bbe9b1dbeb10f10f3a136d01d2a5b6f53d
+  metadata.gz: 62cfaa4fff44e045a7adf83a7bc3ccb29db84812
+  data.tar.gz: 8473e5cb764e1474f2aa541b195afd22ca123a84
 SHA512:
-  metadata.gz: 38ff1874cd0fea432da1e3cc0303c674bef437c689cb4bf47e7f43d1072aaf6569dbc53727fdfffd9e7479f18912a86e49fd1c6f9350ac09c5a5d2acfa53aa7d
-  data.tar.gz: a1c512f261e1b2b1cb704a5b6ba41b1d172b9191b17c04a031e54366e664f1485ecf1e71c2a6993b7cfa49b35ffd383c571e464703fb1eccdcfb088bad3c2c19
+  metadata.gz: 2e2911b2ab856cc6ce7514f289a02cf1fe3e29e9e4a546aca1a11a97da8c4f4b9a2518074ed8b1b4148ffde164c62a970d36f42cd1c6e56b9c49021f513b818f
+  data.tar.gz: f8664d3e2e048c9026865929c1c39375242b3f05d669b6b14212442f3ffaeb0aeca03a8a80091c42e1b59d6e305d7622b5ffb8eca516f083a4af991aaeaf8ef7

data/CHANGELOG.md

@@ -1,3 +1,6 @@
+# 0.0.8 / 2014-04-09
+* [FEATURE] Allow degree of set membership to be included when adding data and incrementally increase scores when adding new data
+
 # 0.0.7 / 2014-04-01
 * [BUGFIX] Cope with sets of empty tags being set in TagSet
 

data/bin/commendo-find-identical-pairs

@@ -0,0 +1,24 @@
+#!/usr/bin/env ruby
+
+content_set_base_redis_key = ARGV[0]
+
+require 'redis'
+require 'commendo'
+
+redis = Redis.new(db: 2)
+#cs = Commendo::ContentSet.new(redis, content_set_base_redis_key)
+
+distribution = {}
+cursor = 0
+begin
+  cursor, keys = redis.scan(cursor.to_i, match: "#{content_set_base_redis_key}*", count: 1000)
+  keys.each do |key|
+    left_resource = key.gsub(/^#{content_set_base_redis_key}/, '')
+    similar = redis.zrange(key, 0, -1, with_scores: true)
+    similar.each do |recommendation|
+      right_resource, score = recommendation
+      puts "#{left_resource}\t#{right_resource}" if score == 1 && left_resource < right_resource
+    end
+  end
+end while cursor.to_i > 0
+

data/bin/commendo-load

@@ -0,0 +1,21 @@
+#!/usr/bin/env ruby
+
+redis_db = ARGV[0].to_i
+content_set_base_redis_key = ARGV[1]
+filename = ARGV[2]
+
+require 'redis'
+require 'commendo'
+
+redis = Redis.new(db: 2)
+cs = Commendo::ContentSet.new(redis, content_set_base_redis_key)
+
+File.open(filename) do |f|
+
+  current_resource = nil
+  current_headings = []
+
+  f.each_line.with_index do |tsv, i|
+    next if i.zero?
+
+    resource, _classified, root, score, heading = tsv.split(/\t/)

data/bin/commendo-recommendations-distribution

@@ -0,0 +1,24 @@
+#!/usr/bin/env ruby
+
+content_set_base_redis_key = ARGV[0]
+
+require 'redis'
+require 'commendo'
+
+redis = Redis.new(db: 2)
+#cs = Commendo::ContentSet.new(redis, content_set_base_redis_key)
+
+distribution = {}
+cursor = 0
+begin
+  cursor, keys = redis.scan(cursor.to_i, match: content_set_base_redis_key, count: 1000)
+  keys.each do |key|
+    count = redis.zcard(key)
+    distribution[count] ||= 0
+    distribution[count] += 1
+  end
+end while cursor.to_i > 0
+
+distribution.each do |score, count|
+  puts "#{score}\t#{count}"
+end

data/bin/commendo-similarity-distribution

@@ -0,0 +1,27 @@
+#!/usr/bin/env ruby
+
+content_set_base_redis_key = ARGV[0]
+
+require 'redis'
+require 'commendo'
+
+redis = Redis.new(db: 2)
+#cs = Commendo::ContentSet.new(redis, content_set_base_redis_key)
+
+distribution = {}
+cursor = 0
+begin
+  cursor, keys = redis.scan(cursor.to_i, match: content_set_base_redis_key, count: 1000)
+  keys.each do |key|
+    similar = redis.zrange(key, 0, -1, with_scores: true)
+    similar.each do |recommendation|
+      resource, score = recommendation
+      distribution[score.round(1)] ||= 0
+      distribution[score.round(1)] += 1
+    end
+  end
+end while cursor.to_i > 0
+
+distribution.each do |score, count|
+  puts "#{score}\t#{count}"
+end

data/lib/commendo/content_set.rb

@@ -9,33 +9,33 @@ module Commendo
     end
 
     def add_by_group(group, *resources)
-      redis.sadd(group_key(group), resources)
       resources.each do |resource|
-        redis.sadd(resource_key(resource), group)
+        if resource.kind_of?(Array)
+          add_single(resource[0], group, resource[1])
+        else
+          add_single(resource, group, 1)
+        end
       end
     end
 
     def add(resource, *groups)
-      redis.sadd(resource_key(resource), groups)
       groups.each do |group|
-        redis.sadd(group_key(group), resource)
+        if group.kind_of?(Array)
+          add_single(resource, group[0], group[1])
+        else
+          add_single(resource, group, 1)
+        end
       end
     end
 
+    def add_single(resource, group, score)
+      redis.zincrby(group_key(group), score, resource)
+      redis.zincrby(resource_key(resource), score, group)
+    end
+
     def add_and_calculate(resource, *groups)
       add(resource, *groups)
-      groups = redis.smembers(resource_key(resource))
-      group_keys = groups.map { |group| group_key(group) }
-      resources = redis.sunion(*group_keys)
-      resources.combination(2) do |l, r|
-        intersect = redis.sinter(resource_key(l), resource_key(r))
-        if (intersect.length > 0)
-          union = redis.sunion(resource_key(l), resource_key(r))
-          jaccard = intersect.length / union.length.to_f
-          redis.zadd(similarity_key(l), jaccard, r)
-          redis.zadd(similarity_key(r), jaccard, l)
-        end
-      end
+      calculate_similarity_for_resource(resource, 0)
     end
 
     def delete(resource)
@@ -48,25 +48,43 @@ module Commendo
       redis.del(resource_key(resource))
     end
 
+    SET_TOO_LARGE_FOR_LUA = 999
+
     def calculate_similarity(threshold = 0)
       #TODO make this use scan for scaling
       keys = redis.keys("#{resource_key_base}:*")
       keys.each_with_index do |key, i|
         yield(key, i, keys.length) if block_given?
-        completed = redis.eval(similarity_lua, keys: [key], argv: [resource_key_base, similar_key_base, group_key_base, threshold])
-        if completed == 999
-          resource = key.gsub(/^#{resource_key_base}:/, '')
-          groups = redis.smembers(resource_key(resource))
-          group_keys = groups.map { |group| group_key(group) }
-          resources = redis.sunion(*group_keys)
-          resources.each do |to_compare|
-            next if resource == to_compare
-            redis.eval(pair_comparison_lua, keys: [key, resource_key(to_compare), similarity_key(resource), similarity_key(to_compare)], argv: [resource, to_compare, threshold])
-          end
+        completed = redis.eval(similarity_lua, keys: [key], argv: [tmp_key_base, resource_key_base, similar_key_base, group_key_base, threshold])
+        if completed == SET_TOO_LARGE_FOR_LUA
+          calculate_similarity_for_key(key, threshold)
         end
       end
     end
 
+    def calculate_similarity_for_key(key, threshold)
+      resource = key.gsub(/^#{resource_key_base}:/, '')
+      calculate_similarity_for_key_resource(key, resource, threshold)
+    end
+
+    def calculate_similarity_for_resource(resource, threshold)
+      key = resource_key(resource)
+      calculate_similarity_for_key_resource(key, resource, threshold)
+    end
+
+    def calculate_similarity_for_key_resource(key, resource, threshold)
+      groups = redis.zrange(resource_key(resource), 0, -1)
+      group_keys = groups.map { |group| group_key(group) }
+      tmp_key = "#{tmp_key_base}:#{SecureRandom.uuid}"
+      redis.zunionstore(tmp_key, group_keys)
+      resources = redis.zrange(tmp_key, 0, -1)
+      redis.del(tmp_key)
+      resources.each do |to_compare|
+        next if resource == to_compare
+        redis.eval(pair_comparison_lua, keys: [key, resource_key(to_compare), similarity_key(resource), similarity_key(to_compare)], argv: [tmp_key_base, resource, to_compare, threshold])
+      end
+    end
+
     def similar_to(resource)
       if resource.kind_of? Array
         keys = resource.map do |res|
@@ -116,6 +134,10 @@ module Commendo
       file.read
     end
 
+    def tmp_key_base
+      "#{key_base}:tmp"
+    end
+
     def similar_key_base
       "#{key_base}:similar"
     end

data/lib/commendo/pair_comparison.lua

@@ -3,9 +3,10 @@ local right_key = KEYS[2]
 local left_similarity_key = KEYS[3]
 local right_similarity_key = KEYS[4]
 
-local left = tonumber(ARGV[1])
-local right = tonumber(ARGV[2])
-local threshold = tonumber(ARGV[3])
+local tmp_key_base = ARGV[1]
+local left = tonumber(ARGV[2])
+local right = tonumber(ARGV[3])
+local threshold = tonumber(ARGV[4])
 
 local function round(num, idp)
   local mult = 10^(idp or 0)
@@ -14,10 +15,31 @@ end
 
 redis.log(redis.LOG_NOTICE, 'Running pair comparison for ' .. left_key .. ' ' .. right_key)
 
-local intersect = table.getn(redis.call('SINTER', left_key, right_key))
-if intersect > 0 then
-    local union = table.getn(redis.call('SUNION', left_key, right_key))
-    local similarity = round(intersect / union, 3)
+
+--TODO change bar
+local tmp_pair_intersect_key = tmp_key_base .. 'bar'
+redis.call('ZINTERSTORE', tmp_pair_intersect_key, 2, left_key, right_key)
+local intersect = redis.call('ZRANGE', tmp_pair_intersect_key, 0, -1, 'WITHSCORES')
+redis.call('DEL', tmp_pair_intersect_key)
+
+if table.getn(intersect) > 0 then
+    local intersect_score = 0
+    for i=1,#intersect,2 do
+        intersect_score = intersect_score + intersect[i+1]
+    end
+
+    --TODO change baz
+    local tmp_pair_union_key = tmp_key_base .. 'baz'
+    redis.call('ZUNIONSTORE', tmp_pair_union_key, 2, left_key, right_key)
+
+    local union = redis.call('ZRANGE', tmp_pair_union_key, 0, -1, 'WITHSCORES')
+    redis.call('DEL', tmp_pair_union_key)
+    local union_score = 0
+    for i=1,#union,2 do
+        union_score = union_score + union[i+1]
+    end
+
+    local similarity = round(intersect_score / union_score, 3)
     if similarity > threshold then
         redis.call('ZADD', left_similarity_key, similarity, right)
         redis.call('ZADD', right_similarity_key, similarity, left)

data/lib/commendo/similarity.lua

@@ -1,8 +1,9 @@
 local resource_key = KEYS[1]
-local resource_key_base = ARGV[1]
-local sim_key_base = ARGV[2]
-local group_key_base = ARGV[3]
-local threshold = tonumber(ARGV[4])
+local tmp_key_base = ARGV[1]
+local resource_key_base = ARGV[2]
+local sim_key_base = ARGV[3]
+local group_key_base = ARGV[4]
+local threshold = tonumber(ARGV[5])
 
 local function round(num, idp)
   local mult = 10^(idp or 0)
@@ -12,7 +13,7 @@ end
 redis.log(redis.LOG_NOTICE, 'Running complete similarity for ' .. resource_key)
 
 local resource = resource_key:gsub('%' .. resource_key_base .. ':', '')
-local groups = redis.call('smembers', resource_key)
+local groups = redis.call('ZRANGE', resource_key, 0, -1)
 
 if table.getn(groups) > 999 then
     redis.log(redis.LOG_NOTICE, 'Complete similarity too large for ' .. resource_key .. ', ' .. table.getn(groups))
@@ -25,24 +26,45 @@ for _,group in ipairs(groups) do
 end
 --redis.log(redis.LOG_NOTICE, 'Found ' .. table.getn(group_keys) .. ' group keys')
 
-local resources = redis.call('sunion', unpack(group_keys))
-
---local resources = redis.call('sunion', unpack(group_keys))
---redis.log(redis.LOG_NOTICE, 'Found ' .. table.getn(resources) .. ' resources')
+--TODO change foo
+local tmp_groups_union_key = tmp_key_base .. 'unionfoo'
+redis.call('ZUNIONSTORE', tmp_groups_union_key, table.getn(group_keys), unpack(group_keys))
+local resources = redis.call('ZRANGE', tmp_groups_union_key, 0, -1)
 
 local previous = 'foo'
 for _,to_compare in ipairs(resources) do
---    redis.log(redis.LOG_NOTICE, 'Comparing ' .. resource .. ' and ' .. to_compare)
+    --redis.log(redis.LOG_NOTICE, 'Comparing ' .. resource .. ' and ' .. to_compare)
     if to_compare ~= previous then
         previous = to_compare
         if resource > to_compare then
---          redis.log(redis.LOG_NOTICE, 'Calculating similarity for ' .. resource .. ' and ' .. to_compare)
-            local intersect = table.getn(redis.call('SINTER', resource_key, resource_key_base .. ':' .. to_compare))
-            if intersect > 0 then
-                local union = table.getn(redis.call('SUNION', resource_key, resource_key_base .. ':' .. to_compare))
-                local similarity = round(intersect / union, 3)
+          --redis.log(redis.LOG_NOTICE, 'Calculating similarity for ' .. resource .. ' and ' .. to_compare)
+
+            --TODO change bar
+            local tmp_pair_intersect_key = tmp_key_base .. 'bar'
+            redis.call('ZINTERSTORE', tmp_pair_intersect_key, 2, resource_key, resource_key_base .. ':' .. to_compare)
+            local intersect = redis.call('ZRANGE', tmp_pair_intersect_key, 0, -1, 'WITHSCORES')
+            redis.call('DEL', tmp_pair_intersect_key)
+
+            if table.getn(intersect) > 0 then
+                local intersect_score = 0
+                for i=1,#intersect,2 do
+                    intersect_score = intersect_score + intersect[i+1]
+                end
+
+                --TODO change baz
+                local tmp_pair_union_key = tmp_key_base .. 'baz'
+                redis.call('ZUNIONSTORE', tmp_pair_union_key, 2, resource_key, resource_key_base .. ':' .. to_compare)
+
+                local union = redis.call('ZRANGE', tmp_pair_union_key, 0, -1, 'WITHSCORES')
+                redis.call('DEL', tmp_pair_union_key)
+                local union_score = 0
+                for i=1,#union,2 do
+                    union_score = union_score + union[i+1]
+                end
+
+                local similarity = round(intersect_score / union_score, 3)
                 if similarity > threshold then
---                  redis.log(redis.LOG_NOTICE, resource .. ' and ' .. to_compare .. ' scored ' .. similarity)
+                  --redis.log(redis.LOG_NOTICE, resource .. ' and ' .. to_compare .. ' scored ' .. similarity)
                     redis.call('ZADD', sim_key_base .. ':' .. resource, similarity, to_compare)
                     redis.call('ZADD', sim_key_base .. ':' .. to_compare, similarity, resource)
                 end
@@ -51,6 +73,12 @@ for _,to_compare in ipairs(resources) do
     end
 end
 
+redis.call('DEL', tmp_groups_union_key)
+
 redis.log(redis.LOG_NOTICE, 'Finished running complete similarity for ' .. resource_key)
 
-return true
+return true
+
+
+
+

data/lib/commendo/version.rb

@@ -1,3 +1,3 @@
 module Commendo
-  VERSION = '0.0.7'
+  VERSION = '0.0.8'
 end

data/test/content_set_test.rb

@@ -16,7 +16,7 @@ module Commendo
       assert_equal 'CommendoTests:similar:resource-1', cs.similarity_key('resource-1')
     end
 
-    def test_stores_sets_by_resource
+    def test_recommends_when_added
       redis = Redis.new(db: 15)
       redis.flushdb
       key_base = 'CommendoTests'
@@ -25,33 +25,89 @@ module Commendo
       cs.add('resource-2', 'group-1')
       cs.add('resource-3', 'group-1', 'group-2')
       cs.add('resource-4', 'group-2')
-      assert redis.sismember("#{key_base}:resources:resource-1", 'group-1')
-      assert redis.sismember("#{key_base}:resources:resource-2", 'group-1')
-      assert redis.sismember("#{key_base}:resources:resource-3", 'group-1')
-      refute redis.sismember("#{key_base}:resources:resource-4", 'group-1')
-
-      assert redis.sismember("#{key_base}:resources:resource-1", 'group-2')
-      refute redis.sismember("#{key_base}:resources:resource-2", 'group-2')
-      assert redis.sismember("#{key_base}:resources:resource-3", 'group-2')
-      assert redis.sismember("#{key_base}:resources:resource-4", 'group-2')
+      cs.calculate_similarity
+      expected = [
+        {resource: 'resource-3', similarity: 1.0},
+        {resource: 'resource-4', similarity: 0.667},
+        {resource: 'resource-2', similarity: 0.667}
+      ]
+      assert_equal expected, cs.similar_to('resource-1')
     end
 
-    def test_stores_sets_by_group
+    def test_recommends_when_added_with_scores
+      redis = Redis.new(db: 15)
+      redis.flushdb
+      key_base = 'CommendoTests'
+      cs = ContentSet.new(redis, key_base)
+      cs.add('resource-1', ['group-1', 2], ['group-2', 2])
+      cs.add('resource-2', ['group-1', 7])
+      cs.add('resource-3', ['group-1', 2], ['group-2', 2])
+      cs.add('resource-4', ['group-2', 3])
+      cs.calculate_similarity
+      expected = [
+        {resource: 'resource-3', similarity: 1.0},
+        {resource: 'resource-2', similarity: 0.818},
+        {resource: 'resource-4', similarity: 0.714}
+      ]
+      assert_equal expected, cs.similar_to('resource-1')
+    end
+
+    def test_recommends_when_extra_scores_added
+      test_recommends_when_added_with_scores
+      redis = Redis.new(db: 15)
+      key_base = 'CommendoTests'
+      cs = ContentSet.new(redis, key_base)
+      cs.add('resource-3', ['group-1', 1], ['group-3', 2])
+      cs.add('resource-4', ['group-2', 1])
+      cs.add_by_group('group-1', ['newource-9', 100], 'resource-2', 'resource-3')
+      cs.add_by_group('group-2', 'resource-1', 'resource-3', 'resource-4')
+      cs.calculate_similarity
+      expected = [
+        {resource: 'newource-9', similarity: 1.0},
+        {resource: 'resource-1', similarity: 0.769},
+        {resource: 'resource-3', similarity: 0.706}
+      ]
+      assert_equal expected, cs.similar_to('resource-2')
+    end
+
+    def test_recommends_when_added_by_group
       redis = Redis.new(db: 15)
       redis.flushdb
       key_base = 'CommendoTests'
       cs = ContentSet.new(redis, key_base)
       cs.add_by_group('group-1', 'resource-1', 'resource-2', 'resource-3')
       cs.add_by_group('group-2', 'resource-1', 'resource-3', 'resource-4')
-      assert redis.sismember("#{key_base}:resources:resource-1", 'group-1')
-      assert redis.sismember("#{key_base}:resources:resource-2", 'group-1')
-      assert redis.sismember("#{key_base}:resources:resource-3", 'group-1')
-      refute redis.sismember("#{key_base}:resources:resource-4", 'group-1')
-
-      assert redis.sismember("#{key_base}:resources:resource-1", 'group-2')
-      refute redis.sismember("#{key_base}:resources:resource-2", 'group-2')
-      assert redis.sismember("#{key_base}:resources:resource-3", 'group-2')
-      assert redis.sismember("#{key_base}:resources:resource-4", 'group-2')
+      cs.calculate_similarity
+      expected = [
+        {resource: 'resource-3', similarity: 1.0},
+        {resource: 'resource-4', similarity: 0.667},
+        {resource: 'resource-2', similarity: 0.667}
+      ]
+      assert_equal expected, cs.similar_to('resource-1')
+    end
+
+    def test_recommends_when_added_by_group_with_scores
+      redis = Redis.new(db: 15)
+      redis.flushdb
+      key_base = 'CommendoTests'
+      cs = ContentSet.new(redis, key_base)
+      cs.add_by_group('group-1', ['resource-1', 2], ['resource-2', 3], ['resource-3', 7])
+      cs.add_by_group('group-2', ['resource-1', 2], ['resource-3', 3], ['resource-4', 5])
+      cs.calculate_similarity
+      expected = [
+        {resource: 'resource-3', similarity: 1.0},
+        {resource: 'resource-4', similarity: 0.778},
+        {resource: 'resource-2', similarity: 0.714}
+      ]
+      assert_equal expected, cs.similar_to('resource-1')
+    end
+
+    def test_recommendations_are_isolated_by_key_base
+      skip
+    end
+
+    def test_recommendations_are_isolated_by_redis_db
+      skip
     end
 
     def test_calculates_similarity_scores
@@ -66,12 +122,12 @@ module Commendo
       end
       cs.calculate_similarity
       expected = [
-        {resource: '9', similarity: 0.5},
-        {resource: '6', similarity: 0.5},
-        {resource: '12', similarity: 0.333},
-        {resource: '3', similarity: 0.25},
-        {resource: '21', similarity: 0.167},
-        {resource: '15', similarity: 0.167}
+        {resource: '9', similarity: 0.667},
+        {resource: '6', similarity: 0.667},
+        {resource: '12', similarity: 0.5},
+        {resource: '3', similarity: 0.4},
+        {resource: '21', similarity: 0.286},
+        {resource: '15', similarity: 0.286}
       ]
       assert_equal expected, cs.similar_to(18)
     end
@@ -88,8 +144,9 @@ module Commendo
       end
       cs.calculate_similarity(0.4)
       expected = [
-        {resource: '9', similarity: 0.5},
-        {resource: '6', similarity: 0.5},
+        {resource: '9', similarity: 0.667},
+        {resource: '6', similarity: 0.667},
+        {resource: '12', similarity: 0.5}
       ]
       assert_equal expected, cs.similar_to(18)
     end
@@ -250,17 +307,17 @@ module Commendo
       end
       cs.calculate_similarity
       expected = [
-        {resource: '18', similarity: 1.333},
-        {resource: '3', similarity: 1.25},
-        {resource: '6', similarity: 0.833},
-        {resource: '12', similarity: 0.7},
-        {resource: '21', similarity: 0.667},
-        {resource: '15', similarity: 0.667},
-        {resource: '9', similarity: 0.533},
-        {resource: '4', similarity: 0.25},
-        {resource: '8', similarity: 0.2},
-        {resource: '16', similarity: 0.167},
-        {resource: '20', similarity: 0.143}
+        {resource: '18', similarity: 1.834},
+        {resource: '3', similarity: 1.734},
+        {resource: '6', similarity: 1.167},
+        {resource: '21', similarity: 1.086},
+        {resource: '15', similarity: 1.086},
+        {resource: '12', similarity: 1.0},
+        {resource: '9', similarity: 0.833},
+        {resource: '4', similarity: 0.4},
+        {resource: '8', similarity: 0.333},
+        {resource: '16', similarity: 0.286},
+        {resource: '20', similarity: 0.25}
       ]
       actual = cs.similar_to([12, 6, 9])
       assert_equal expected, actual

data/test/weighted_group_test.rb

@@ -19,9 +19,9 @@ module Commendo
       @cs3 = ContentSet.new(@redis, 'CommendoTests:ContentSet3', @tag_set)
       (3..23).each do |group|
         (3..23).each do |res|
-          @cs1.add_by_group(group, res) if (res % group == 0) && (res % 2 == 0)
-          @cs2.add_by_group(group, res) if (res % group == 0) && (res % 3 == 0)
-          @cs3.add_by_group(group, res) if (res % group == 0) && (res % 6 == 0)
+          @cs1.add_by_group(group, res) if res.modulo(group).zero? && res.modulo(2).zero?
+          @cs2.add_by_group(group, res) if res.modulo(group).zero? && res.modulo(3).zero?
+          @cs3.add_by_group(group, res) if res.modulo(group).zero? && res.modulo(6).zero?
           @tag_set.add(res, 'mod3') if res.modulo(3).zero?
           @tag_set.add(res, 'mod4') if res.modulo(4).zero?
           @tag_set.add(res, 'mod5') if res.modulo(5).zero?
@@ -40,12 +40,12 @@ module Commendo
         {cs: @cs3, weight: 100.0}
       )
       expected = [
-        {resource: '6', similarity: 55.5},
-        {resource: '12', similarity: 36.963},
-        {resource: '9', similarity: 5.0},
-        {resource: '3', similarity: 2.5},
-        {resource: '21', similarity: 1.67},
-        {resource: '15', similarity: 1.67}
+        {resource: '6', similarity: 74.037},
+        {resource: '12', similarity: 55.5},
+        {resource: '9', similarity: 6.67},
+        {resource: '3', similarity: 4.0},
+        {resource: '21', similarity: 2.86},
+        {resource: '15', similarity: 2.86}
       ]
       assert_equal expected, weighted_group.similar_to(18)
     end
@@ -58,7 +58,7 @@ module Commendo
         {cs: @cs2, weight: 10.0},
         {cs: @cs3, weight: 100.0}
       )
-      expected = [{resource: '15', similarity: 1.67}]
+      expected = [{resource: '15', similarity: 2.86}]
       weighted_group.tag_set = @tag_set
       assert_equal expected, weighted_group.filtered_similar_to(18, include: ['mod5'])
     end
@@ -72,10 +72,10 @@ module Commendo
         {cs: @cs3, weight: 100.0}
       )
       expected = [
-        {resource: '6', similarity: 55.5},
-        {resource: '12', similarity: 36.963},
-        {resource: '9', similarity: 5.0},
-        {resource: '3', similarity: 2.5}
+        {resource: '6', similarity: 74.037},
+        {resource: '12', similarity: 55.5},
+        {resource: '9', similarity: 6.67},
+        {resource: '3', similarity: 4.0}
       ]
       weighted_group.tag_set = @tag_set
       assert_equal expected, weighted_group.filtered_similar_to(18, exclude: ['mod5', 'mod7'])
@@ -90,9 +90,9 @@ module Commendo
         {cs: @cs3, weight: 1.0}
       )
       expected = [
-        {resource: '16', similarity: 66.7},
-        {resource: '4', similarity: 50.0},
-        {resource: '12', similarity: 20.0}
+        {resource: '16', similarity: 80.0},
+        {resource: '4', similarity: 66.7},
+        {resource: '12', similarity: 33.3}
       ]
       weighted_group.tag_set = @tag_set
       assert_equal expected, weighted_group.filtered_similar_to(8, include: ['mod4'], exclude: ['mod5'])
@@ -107,16 +107,16 @@ module Commendo
         {cs: @cs3, weight: 1.0}
       )
       expected = [
-        {resource: '12', similarity: 83.0},
-        {resource: '18', similarity: 58.0},
-        {resource: '8', similarity: 50.0},
-        {resource: '16', similarity: 33.3},
-        {resource: '20', similarity: 25.0},
-        {resource: '9', similarity: 8.33},
-        {resource: '21', similarity: 5.83},
-        {resource: '15', similarity: 5.83},
-        {resource: '6', similarity: 5.0},
-        {resource: '3', similarity: 5.0}
+        {resource: '12', similarity: 118.037},
+        {resource: '18', similarity: 78.037},
+        {resource: '8', similarity: 66.7},
+        {resource: '16', similarity: 50.0},
+        {resource: '20', similarity: 40.0},
+        {resource: '9', similarity: 11.67},
+        {resource: '21', similarity: 9.0},
+        {resource: '15', similarity: 9.0},
+        {resource: '6', similarity: 6.67},
+        {resource: '3', similarity: 6.67}
       ]
       weighted_group.tag_set = @tag_set
       assert_equal expected, weighted_group.similar_to([3,4,5,6,7])
@@ -131,16 +131,9 @@ module Commendo
         {cs: @cs3, weight: 1.0}
       )
       expected = [
-        {resource: '12', similarity: 83.0},
-        #{resource: '18', similarity: 58.0},
-        {resource: '8', similarity: 50.0},
-        {resource: '16', similarity: 33.3},
-        #{resource: '20', similarity: 25.0},
-        #{resource: '9', similarity: 8.33},
-        #{resource: '21', similarity: 5.83},
-        #{resource: '15', similarity: 5.83},
-        #{resource: '6', similarity: 5.0},
-        #{resource: '3', similarity: 5.0}
+        {resource: '12', similarity: 118.037},
+        {resource: '8', similarity: 66.7},
+        {resource: '16', similarity: 50.0},
       ]
       weighted_group.tag_set = @tag_set
       assert_equal expected, weighted_group.filtered_similar_to([3,4,5,6,7], include: ['mod4'], exclude: ['mod5'])

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: commendo
 version: !ruby/object:Gem::Version
-  version: 0.0.7
+  version: 0.0.8
 platform: ruby
 authors:
 - Rob Styles
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-04-01 00:00:00.000000000 Z
+date: 2014-04-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: redis
@@ -97,7 +97,11 @@ dependencies:
 description: A Jaccard-similarity recommender using Redis sets
 email:
 - rob.styles@dynamicorange.com
-executables: []
+executables:
+- commendo-find-identical-pairs
+- commendo-load
+- commendo-recommendations-distribution
+- commendo-similarity-distribution
 extensions: []
 extra_rdoc_files: []
 files:
@@ -107,6 +111,10 @@ files:
 - LICENSE.txt
 - README.md
 - Rakefile
+- bin/commendo-find-identical-pairs
+- bin/commendo-load
+- bin/commendo-recommendations-distribution
+- bin/commendo-similarity-distribution
 - commendo.gemspec
 - lib/commendo.rb
 - lib/commendo/content_set.rb