グカネータ / 質問パターン見直し (#41) (#365)

Reviewed-on: #365
Co-authored-by: miteruzo <miteruzo@naver.com>
Co-committed-by: miteruzo <miteruzo@naver.com>

backend/app/controllers/gekanator_games_controller.rb

@@ -27,26 +27,20 @@ class GekanatorGamesController < ApplicationController
     game = GekanatorGame.find_by(id: params[:id])
     return head :not_found unless game
 
-    asked_ids = Array(game.answers).filter_map { |answer| answer_question_id(answer) }
-    existing_example_ids =
-        GekanatorQuestionExample.where(post_id: game.correct_post_id)
-                                .select(:gekanator_question_id)
-    # Direct examples only for now; post_similarity-based expansion is deferred.
     questions =
         GekanatorQuestion
         .accepted
+        .includes(:gekanator_question_examples)
         .where(kind: 'post_similarity', source: 'user_suggested')
-        .where.not(id: existing_example_ids)
-        .order(priority_weight: :desc, id: :asc)
+        .to_a
+
+    selected = weighted_sample_questions(
+        questions,
+        post_id: game.correct_post_id,
+        limit: 2)
 
     render json: {
-      questions: questions.filter_map { |question|
-        json = extra_question_json(question)
-        next if asked_ids.include?(json[:id].to_s)
-        next if asked_ids.include?("post-similarity:#{ json[:id] }")
-
-        json
-      }.first(2)
+      questions: selected.map { |question| extra_question_json(question) }
     }
   end
 
@@ -84,11 +78,10 @@ class GekanatorGamesController < ApplicationController
                 gekanator_question: question,
                 post: game.correct_post,
                 user: current_user)
-        example.assign_attributes(
-            gekanator_game: game,
+        example.record_answer!(
             answer: item[:answer],
             source: 'post_game_extra',
-            weight: 1.0)
+            gekanator_game: game)
         example.save!
       end
     end
@@ -107,12 +100,41 @@ class GekanatorGamesController < ApplicationController
     }
   end
 
-  def answer_question_id answer
-    value = if answer.is_a?(Hash)
-      answer['question_id'].presence || answer[:question_id].presence ||
-        answer['questionId'].presence || answer[:questionId].presence
+  def weighted_sample_questions questions, post_id:, limit:
+    remaining = questions.uniq(&:id)
+    selected = []
+
+    while selected.length < limit && remaining.any?
+      weighted =
+          remaining.map { |question|
+            [question, selection_weight_for(question, post_id: post_id)]
+          }
+      total_weight = weighted.sum { |_question, weight| weight }
+      break if total_weight <= 0
+
+      target = rand * total_weight
+      cumulative = 0.0
+      chosen =
+          weighted.find do |_question, weight|
+            cumulative += weight
+            cumulative >= target
+          end&.first || weighted.first.first
+
+      selected << chosen
+      remaining.reject! { |question| question.id == chosen.id }
     end
 
-    value&.to_s
+    selected
+  end
+
+  def selection_weight_for question, post_id:
+    sample_count =
+        question.gekanator_question_examples.sum { |example|
+          next 0 unless example.post_id == post_id
+
+          example.sample_count.presence || 1
+        }
+
+    question.priority_weight.to_f / (1.0 + sample_count * 0.15)
   end
 end

backend/app/models/gekanator_question_example.rb

@@ -1,5 +1,6 @@
 class GekanatorQuestionExample < ApplicationRecord
   ANSWERS = GekanatorQuestionSuggestion::ANSWERS
+  NON_UNKNOWN_ANSWERS = ANSWERS - ['unknown']
   SOURCES = ['initial_suggestion', 'post_game_extra'].freeze
 
   belongs_to :gekanator_question
@@ -8,10 +9,89 @@ class GekanatorQuestionExample < ApplicationRecord
   belongs_to :gekanator_game, optional: true
 
   validates :answer, presence: true, inclusion: { in: ANSWERS }
+  validates :answer_counts, presence: true
+  validates :sample_count,
+            presence: true,
+            numericality: {
+              only_integer: true,
+              greater_than: 0
+            }
   validates :source, presence: true, inclusion: { in: SOURCES }
   validates :weight,
             presence: true,
             numericality: {
               greater_than: 0
             }
+
+  before_validation :normalize_learning_state
+
+  def record_answer!(answer:, source:, gekanator_game: nil)
+    answer = answer.to_s
+    raise ArgumentError, 'invalid answer' unless ANSWERS.include?(answer)
+
+    counts = normalized_answer_counts
+    counts[answer] += 1
+
+    self.answer_counts = counts
+    self.sample_count = counts.values.sum
+    self.gekanator_game = gekanator_game if gekanator_game.present?
+    self.source = source if new_record?
+
+    apply_aggregated_answer!(preferred_answer: answer)
+    self
+  end
+
+  private
+
+  def normalize_learning_state
+    counts = normalized_answer_counts
+
+    if counts.values.sum.zero? && answer.present?
+      counts[answer] = 1
+    end
+
+    self.answer_counts = counts
+    self.sample_count = counts.values.sum
+
+    apply_aggregated_answer!
+  end
+
+  def apply_aggregated_answer!(preferred_answer: nil)
+    counts = normalized_answer_counts
+    known_counts = counts.slice(*NON_UNKNOWN_ANSWERS)
+    known_total = known_counts.values.sum
+
+    if known_total.zero?
+      self.answer = 'unknown'
+      self.weight = 0.1
+      return
+    else
+      max_count = known_counts.values.max
+      candidates = known_counts.select { |_answer, count| count == max_count }.keys
+      self.answer =
+          if preferred_answer.present? && candidates.include?(preferred_answer)
+            preferred_answer
+          elsif answer.present? && candidates.include?(answer)
+            answer
+          else
+            candidates.first
+          end
+    end
+
+    consensus = max_count.to_f / known_total
+    self.weight = Math.sqrt(known_total) * consensus
+  end
+
+  def normalized_answer_counts
+    base = ANSWERS.index_with(0)
+
+    answer_counts.to_h.each do |key, value|
+      answer_key = key.to_s
+      next unless ANSWERS.include?(answer_key)
+
+      base[answer_key] = value.to_i
+    end
+
+    base
+  end
 end

backend/app/services/gekanator/question_suggestion_promoter.rb

@@ -26,13 +26,16 @@ module Gekanator
             },
             gekanator_question_suggestion: suggestion,
             created_by: user)
-        GekanatorQuestionExample.create!(
-            gekanator_question: question,
-            post: suggestion.gekanator_game.correct_post,
-            user: user,
-            gekanator_game: suggestion.gekanator_game,
+        example =
+            GekanatorQuestionExample.new(
+                gekanator_question: question,
+                post: suggestion.gekanator_game.correct_post,
+                user: user)
+        example.record_answer!(
             answer: suggestion.answer,
-            source: 'initial_suggestion')
+            source: 'initial_suggestion',
+            gekanator_game: suggestion.gekanator_game)
+        example.save!
         suggestion.update!(processed: true)
         question
       end

backend/db/migrate/20260612000000_add_answer_statistics_to_gekanator_question_examples.rb

@@ -0,0 +1,40 @@
+class AddAnswerStatisticsToGekanatorQuestionExamples < ActiveRecord::Migration[8.0]
+  class MigrationGekanatorQuestionExample < ApplicationRecord
+    self.table_name = 'gekanator_question_examples'
+  end
+
+  def up
+    add_column :gekanator_question_examples,
+               :answer_counts,
+               :json,
+               null: true
+    add_column :gekanator_question_examples,
+               :sample_count,
+               :integer,
+               null: false,
+               default: 1
+
+    MigrationGekanatorQuestionExample.reset_column_information
+    MigrationGekanatorQuestionExample.find_each do |example|
+      counts = {
+        'yes' => 0,
+        'no' => 0,
+        'partial' => 0,
+        'probably_no' => 0,
+        'unknown' => 0
+      }
+      counts[example.answer] = 1 if counts.key?(example.answer)
+
+      example.update_columns(
+        answer_counts: counts,
+        sample_count: 1)
+    end
+
+    change_column_null :gekanator_question_examples, :answer_counts, false
+  end
+
+  def down
+    remove_column :gekanator_question_examples, :sample_count
+    remove_column :gekanator_question_examples, :answer_counts
+  end
+end

backend/db/schema.rb

@@ -10,7 +10,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema[8.0].define(version: 2026_06_10_000000) do
+ActiveRecord::Schema[8.0].define(version: 2026_06_12_000000) do
   create_table "active_storage_attachments", charset: "utf8mb4", collation: "utf8mb4_0900_ai_ci", force: :cascade do |t|
     t.string "name", null: false
     t.string "record_type", null: false
@@ -85,6 +85,8 @@ ActiveRecord::Schema[8.0].define(version: 2026_06_10_000000) do
     t.float "weight", default: 1.0, null: false
     t.datetime "created_at", null: false
     t.datetime "updated_at", null: false
+    t.json "answer_counts", null: false
+    t.integer "sample_count", default: 1, null: false
     t.index ["gekanator_game_id"], name: "index_gekanator_question_examples_on_gekanator_game_id"
     t.index ["gekanator_question_id", "post_id", "user_id"], name: "idx_gekanator_question_examples_on_question_post_user", unique: true
     t.index ["gekanator_question_id"], name: "index_gekanator_question_examples_on_gekanator_question_id"

backend/spec/requests/gekanator_learning_spec.rb

@@ -275,7 +275,7 @@ RSpec.describe 'Gekanator learning API', type: :request do
   end
 
   describe 'GET /gekanator/games/:id/extra_questions' do
-    it 'returns at most two accepted user_suggested post_similarity questions' do
+    it 'returns at most two accepted user_suggested post_similarity questions without duplicates' do
       sign_in_as admin
 
       low = create_post_similarity_question!(
@@ -295,15 +295,14 @@ RSpec.describe 'Gekanator learning API', type: :request do
 
       expect(response).to have_http_status(:ok)
       expect(json['questions'].length).to eq(2)
-      expect(json['questions'].map { _1['id'] }).to eq([high.id, middle.id])
-      expect(json['questions'].map { _1['id'] }).not_to include(low.id)
+      expect(json['questions'].map { _1['id'] }.uniq.length).to eq(2)
+      expect(json['questions'].map { _1['id'] }).to all(be_in([low.id, high.id, middle.id]))
     end
 
-    it 'does not return questions that already have an example for the correct post' do
+    it 'can return questions that already have an example for the correct post' do
       sign_in_as admin
 
       existing = create_post_similarity_question!(text: 'already learned?')
-      fresh = create_post_similarity_question!(text: 'fresh?')
 
       GekanatorQuestionExample.create!(
         gekanator_question: existing,
@@ -317,15 +316,13 @@ RSpec.describe 'Gekanator learning API', type: :request do
       get "/gekanator/games/#{game.id}/extra_questions"
 
       expect(response).to have_http_status(:ok)
-      expect(json['questions'].map { _1['id'] }).to include(fresh.id)
-      expect(json['questions'].map { _1['id'] }).not_to include(existing.id)
+      expect(json['questions'].map { _1['id'] }).to include(existing.id)
     end
 
-    it 'does not return questions already asked in the game using snake_case question_id' do
+    it 'can return questions already asked in the game using snake_case question_id' do
       sign_in_as admin
 
       asked = create_post_similarity_question!(text: 'already asked?')
-      fresh = create_post_similarity_question!(text: 'fresh?')
       game.update!(
         answers: [
           {
@@ -338,15 +335,13 @@ RSpec.describe 'Gekanator learning API', type: :request do
       get "/gekanator/games/#{game.id}/extra_questions"
 
       expect(response).to have_http_status(:ok)
-      expect(json['questions'].map { _1['id'] }).to include(fresh.id)
-      expect(json['questions'].map { _1['id'] }).not_to include(asked.id)
+      expect(json['questions'].map { _1['id'] }).to include(asked.id)
     end
 
-    it 'does not return questions already asked in the game using camelCase questionId' do
+    it 'can return questions already asked in the game using camelCase questionId' do
       sign_in_as admin
 
       asked = create_post_similarity_question!(text: 'already asked?')
-      fresh = create_post_similarity_question!(text: 'fresh?')
       game.update!(
         answers: [
           {
@@ -359,8 +354,7 @@ RSpec.describe 'Gekanator learning API', type: :request do
       get "/gekanator/games/#{game.id}/extra_questions"
 
       expect(response).to have_http_status(:ok)
-      expect(json['questions'].map { _1['id'] }).to include(fresh.id)
-      expect(json['questions'].map { _1['id'] }).not_to include(asked.id)
+      expect(json['questions'].map { _1['id'] }).to include(asked.id)
     end
 
     it 'does not return non-accepted, non-user_suggested, or non-post_similarity questions' do
@@ -584,5 +578,35 @@ RSpec.describe 'Gekanator learning API', type: :request do
         other_post.id.to_s => 'no'
       )
     end
+
+    it 'normalizes legacy title length questions' do
+      sign_in_as admin
+
+      GekanatorQuestion.create!(
+        text: '題名が長めの投稿?',
+        kind: 'title',
+        source: 'admin_curated',
+        status: 'accepted',
+        priority_weight: 1.0,
+        condition: {
+          type: 'title-length-greater-than',
+          length: 20
+        },
+        created_by: admin
+      )
+
+      get '/gekanator/questions'
+
+      expect(response).to have_http_status(:ok)
+      question_json = json['questions'].find { _1['id'] == 'title:length-at-least:21' }
+      expect(question_json).to include(
+        'text' => 'タイトルは 21 文字以上?',
+        'kind' => 'title'
+      )
+      expect(question_json['condition']).to include(
+        'type' => 'title-length-at-least',
+        'length' => 21
+      )
+    end
   end
 end