#376

backend/app/controllers/gekanator_games_controller.rb

@@ -50,7 +50,7 @@ class GekanatorGamesController < ApplicationController
             questions,
             post_id: game.correct_post_id,
             user: current_user,
-            limit: 2)
+            limit: 6)
 
     render json: {
       questions: selected.map { |question| extra_question_json(question) }

backend/spec/requests/gekanator_learning_spec.rb

@@ -475,28 +475,59 @@ 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 without duplicates' do
+    it 'returns at most six accepted user_suggested post_similarity questions without duplicates' do
       sign_in_as admin
 
+      lowest = create_post_similarity_question!(
+        text: 'lowest?',
+        priority_weight: 0.5
+      )
       low = create_post_similarity_question!(
         text: 'low?',
         priority_weight: 1.0
       )
-      high = create_post_similarity_question!(
-        text: 'high?',
-        priority_weight: 3.0
-      )
       middle = create_post_similarity_question!(
         text: 'middle?',
+        priority_weight: 1.5
+      )
+      medium_high = create_post_similarity_question!(
+        text: 'medium high?',
         priority_weight: 2.0
       )
+      high = create_post_similarity_question!(
+        text: 'high?',
+        priority_weight: 2.5
+      )
+      higher = create_post_similarity_question!(
+        text: 'higher?',
+        priority_weight: 2.8
+      )
+      highest = create_post_similarity_question!(
+        text: 'highest?',
+        priority_weight: 3.0
+      )
+      overflow = create_post_similarity_question!(
+        text: 'overflow?',
+        priority_weight: 2.2
+      )
 
       get "/gekanator/games/#{game.id}/extra_questions"
 
       expect(response).to have_http_status(:ok)
-      expect(json['questions'].length).to eq(2)
-      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]))
+      expect(json['questions'].length).to eq(6)
+      expect(json['questions'].map { _1['id'] }.uniq.length).to eq(6)
+      expect(json['questions'].map { _1['id'] }).to all(
+        be_in([
+          lowest.id,
+          low.id,
+          middle.id,
+          medium_high.id,
+          high.id,
+          higher.id,
+          highest.id,
+          overflow.id,
+        ])
+      )
     end
 
     it 'can return questions that already have an example for the correct post' do

frontend/src/lib/gekanator.test.ts

@@ -4,6 +4,7 @@ import { apiPost } from '@/lib/api'
 import {
   buildGekanatorQuestions,
   expectedAnswerForQuestion,
+  learnedSemanticSideForPost,
   questionIdForCondition,
   restoreGekanatorQuestion,
   saveGekanatorExtraQuestionAnswers,
@@ -188,6 +189,33 @@ describe('expectedAnswerForQuestion', () => {
   })
 })
 
+describe('learnedSemanticSideForPost', () => {
+  it('classifies post_similarity examples as positive, negative, or unknown', () => {
+    const question: StoredGekanatorQuestion = {
+      id: 'post-similarity:10',
+      text: '喜多ちゃんが泣いてる?',
+      kind: 'post_similarity',
+      source: 'user_suggested',
+      priorityWeight: 1.2,
+      condition: {
+        type: 'post-similarity',
+        postId: 123,
+        answer: 'partial',
+        threshold: 0.65,
+      },
+      exampleAnswers: {
+        1: 'yes',
+        2: 'probably_no',
+      },
+    }
+
+    expect(learnedSemanticSideForPost(question, post({ id: 1 }))).toBe('positive')
+    expect(learnedSemanticSideForPost(question, post({ id: 2 }))).toBe('negative')
+    expect(learnedSemanticSideForPost(question, post({ id: 3 }))).toBe('unknown')
+    expect(learnedSemanticSideForPost(question, post({ id: 123 }))).toBe('positive')
+  })
+})
+
 describe('restoreGekanatorQuestion', () => {
   it('uses default source and priority weight when omitted', () => {
     const question = restoreGekanatorQuestion({
@@ -248,7 +276,7 @@ describe('restoreGekanatorQuestion', () => {
     })
 
     expect(question.test(post({ id: 1 }))).toBe(true)
-    expect(question.test(post({ id: 2 }))).toBe(false)
+    expect(question.test(post({ id: 2 }))).toBe(true)
   })
 
   it('normalizes legacy title-length-greater-than questions', () => {

frontend/src/lib/gekanator.ts

@@ -9,11 +9,24 @@ export type GekanatorAnswerValue =
     | 'probably_no'
     | 'unknown'
 
+export type LearnedSemanticSide =
+    | 'positive'
+    | 'negative'
+    | 'unknown'
+
+export type GekanatorQuestionPurpose =
+    | 'effective_user_suggested'
+    | 'learning_user_suggested'
+    | 'normal'
+
 export type GekanatorAnswerLog = {
     questionId:       string
     questionText:     string
     questionCondition?:  GekanatorQuestionCondition
     questionMode?:    'normal' | 'winning_run'
+    questionPurpose?: GekanatorQuestionPurpose
+    effectiveQuestion?: boolean
+    learningQuestion?: boolean
     answer:           GekanatorAnswerValue
     originalAnswer:   GekanatorAnswerValue }
 
@@ -163,6 +176,26 @@ const directExampleAnswerFor = (
   return null
 }
 
+
+export const isLearnedSemanticQuestion = (
+    question:  StoredGekanatorQuestion | GekanatorQuestion,
+): boolean =>
+    question.kind === 'post_similarity'
+    && question.source === 'user_suggested'
+
+
+export const learnedSemanticSideForAnswer = (
+    answer:  GekanatorAnswerValue | null,
+): LearnedSemanticSide => {
+  if (answer === 'yes' || answer === 'partial')
+    return 'positive'
+
+  if (answer === 'no' || answer === 'probably_no')
+    return 'negative'
+
+  return 'unknown'
+}
+
 const countBy = <T extends string | number> (values: T[]): Map<T, number> => {
   const counts = new Map<T, number> ()
   values.forEach (value => counts.set (value, (counts.get (value) ?? 0) + 1))
@@ -285,8 +318,8 @@ const questionMatches = (
 ): boolean => {
   const directAnswer = directExampleAnswerFor (question, post)
   if (directAnswer)
-    return question.condition.type === 'post-similarity'
-	? directAnswer === question.condition.answer
+    return question.kind === 'post_similarity'
+	? learnedSemanticSideForAnswer (directAnswer) === 'positive'
 	: directAnswer === 'yes'
 
   switch (question.condition.type)
@@ -328,6 +361,11 @@ export const expectedAnswerForQuestion = (
 
   switch (question.condition.type)
   {
+  case 'post-similarity':
+    if (question.condition.postId === post.id)
+      return question.condition.answer
+
+    return null
   case 'tag':
   case 'source':
   case 'original-year':
@@ -338,12 +376,17 @@ export const expectedAnswerForQuestion = (
   case 'title-has-ascii':
   case 'title-contains':
     return questionMatches (post, question) ? 'yes' : 'no'
-  case 'post-similarity':
-    return null
   }
 }
 
 
+export const learnedSemanticSideForPost = (
+    question:  StoredGekanatorQuestion | GekanatorQuestion | undefined,
+    post:      Post | null,
+): LearnedSemanticSide =>
+    learnedSemanticSideForAnswer (expectedAnswerForQuestion (question, post))
+
+
 export const restoreGekanatorQuestion = (
     question:  StoredGekanatorQuestion,
 ): GekanatorQuestion => {
@@ -423,15 +466,15 @@ export const buildGekanatorQuestions = (
   const originalYears = countBy (
       posts
       .map (originalYearOf)
-      .filter ((year): year is number => year !== null))
+      .filter ((year): year is number => year != null))
   const originalMonths = countBy (
       posts
       .map (originalMonthOf)
-      .filter ((month): month is number => month !== null))
+      .filter ((month): month is number => month != null))
   const originalMonthDays = countBy (
       posts
       .map (originalMonthDayOf)
-      .filter ((monthDay): monthDay is string => monthDay !== null))
+      .filter ((monthDay): monthDay is string => monthDay != null))
   const titleLengthMedian = median (posts.map (post => post.title?.length ?? 0))
   const titleWordCounts =
       includeTitleContains

frontend/src/lib/gekanatorCandidateRecovery.test.ts

@@ -11,6 +11,7 @@ import type {
     GekanatorAnswerValue,
     GekanatorQuestion,
 } from '@/lib/gekanator'
+import type { RecoveredCandidateState } from '@/lib/gekanatorCandidateRecovery'
 import type { Post } from '@/types'
 
 
@@ -78,6 +79,15 @@ const answer = (
 })
 
 
+const recoveredState = (
+    answerCountAtRecovery:  number,
+    scoreAtRecovery = 0,
+): RecoveredCandidateState => ({
+  answerCountAtRecovery,
+  scoreAtRecovery,
+})
+
+
 describe('candidatePostsFor', () => {
   it('does not hard-filter semantic post_similarity answers', () => {
     const posts = [post (1), post (2), post (3)]
@@ -99,8 +109,8 @@ describe('candidatePostsFor', () => {
 	softenedQuestionIds: new Set (),
 	rejectedPostIds: new Set (),
 	recoveredCandidatePosts: new Map ([
-	    [1, 1],
-	    [3, 1],
+	    [1, recoveredState (1)],
+	    [3, recoveredState (1)],
 	]) })
 
     expect(candidates.map (candidate => candidate.id)).toEqual ([1, 2, 3])
@@ -122,8 +132,8 @@ describe('candidatePostsFor', () => {
 	softenedQuestionIds: new Set (),
 	rejectedPostIds: new Set (),
 	recoveredCandidatePosts: new Map ([
-	    [1, 1],
-	    [3, 1],
+	    [1, recoveredState (1)],
+	    [3, recoveredState (1)],
 	]) })
 
     expect(candidates.map (candidate => candidate.id)).toEqual ([3])
@@ -142,7 +152,7 @@ describe('candidatePostsFor', () => {
 	answers: [answer (question, 'yes')],
 	softenedQuestionIds: new Set (),
 	rejectedPostIds: new Set ([1]),
-	recoveredCandidatePosts: new Map ([[1, 1]]) })
+	recoveredCandidatePosts: new Map ([[1, recoveredState (1)]]) })
 
     expect(candidates.map (candidate => candidate.id)).toEqual ([2])
   })
@@ -209,7 +219,7 @@ describe('recoverCandidatePosts', () => {
 	posts,
 	scores,
 	rejectedPostIds: new Set ([10]),
-	recoveredCandidatePosts: new Map ([[8, 1]]),
+	recoveredCandidatePosts: new Map ([[8, recoveredState (1, 8)]]),
 	eligiblePostIds: new Set ([9]),
 	answerCountAtRecovery: 2,
 	recoveryStepCount: 0,
@@ -218,7 +228,10 @@ describe('recoverCandidatePosts', () => {
     expect(recovered?.recoveryStepCount).toBe (1)
     expect([...(recovered?.recoveredCandidatePosts.keys () ?? [])])
       .toEqual ([8, 7, 6, 5, 4])
-    expect(recovered?.recoveredCandidatePosts.get (7)).toBe (2)
+    expect(recovered?.recoveredCandidatePosts.get (7)).toEqual ({
+	answerCountAtRecovery: 2,
+	scoreAtRecovery: 7,
+    })
   })
 
   it('does not add posts when recovered and eligible candidates already hit the target', () => {
@@ -230,9 +243,9 @@ describe('recoverCandidatePosts', () => {
 	scores,
 	rejectedPostIds: new Set (),
 	recoveredCandidatePosts: new Map ([
-	    [1, 1],
-	    [2, 1],
-	    [3, 1],
+	    [1, recoveredState (1, 1)],
+	    [2, recoveredState (1, 2)],
+	    [3, recoveredState (1, 3)],
 	]),
 	eligiblePostIds: new Set ([4, 5, 6]),
 	answerCountAtRecovery: 2,

frontend/src/lib/gekanatorCandidateRecovery.ts

@@ -1,15 +1,21 @@
-import { expectedAnswerForQuestion } from '@/lib/gekanator'
+import { isLearnedSemanticQuestion,
+	 learnedSemanticSideForPost } from '@/lib/gekanator'
 
 import type { GekanatorAnswerLog, GekanatorAnswerValue, GekanatorQuestion } from '@/lib/gekanator'
 import type { Post } from '@/types'
 
 export type RecoveredCandidatePost = {
     postId:                number
-    answerCountAtRecovery: number }
+    answerCountAtRecovery: number
+    scoreAtRecovery:       number }
+
+export type RecoveredCandidateState = {
+    answerCountAtRecovery: number
+    scoreAtRecovery:       number }
 
 
-const questionIsFactLikeForHardFiltering = (question:  GekanatorQuestion): boolean =>
-    !(question.kind === 'post_similarity'
+const questionSupportsAnswerBasedHardFiltering = (question:  GekanatorQuestion): boolean =>
+    !(isLearnedSemanticQuestion (question)
       || (question.kind === 'tag'
 	  && question.condition.type === 'tag'
 	  && !(question.condition.key.startsWith ('nico:'))))
@@ -26,7 +32,7 @@ export const candidatePostsFor = (
 				   answers:                 GekanatorAnswerLog[]
 				   softenedQuestionIds:     Set<string>
 				   rejectedPostIds:         Set<number>
-				   recoveredCandidatePosts: Map<number, number> },
+				   recoveredCandidatePosts: Map<number, RecoveredCandidateState> },
 ): Post[] => {
   const questionById = new Map (questions.map (question => [question.id, question]))
 
@@ -34,10 +40,10 @@ export const candidatePostsFor = (
     if (rejectedPostIds.has (post.id))
       return false
 
-    const answerCountAtRecovery = recoveredCandidatePosts.get (post.id)
+    const recoveredCandidate = recoveredCandidatePosts.get (post.id)
 
     return answers.every ((answer, index) => {
-      if (answerCountAtRecovery != null && index < answerCountAtRecovery)
+      if (recoveredCandidate != null && index < recoveredCandidate.answerCountAtRecovery)
 	return true
 
       if (softenedQuestionIds.has (answer.questionId))
@@ -46,7 +52,7 @@ export const candidatePostsFor = (
       const question = questionById.get (answer.questionId)
       if (!(question))
 	return true
-      if (!(questionIsFactLikeForHardFiltering (question)))
+      if (!(questionSupportsAnswerBasedHardFiltering (question)))
 	return true
 
       switch (answer.answer)
@@ -54,8 +60,10 @@ export const candidatePostsFor = (
       case 'yes':
       case 'no':
 	{
-	  const expected = expectedAnswerForQuestion (question, post)
-	  return expected === null || expected === 'unknown' || expected === answer.answer
+	  const expected = learnedSemanticSideForPost (question, post)
+	  return expected === 'unknown'
+	      || (answer.answer === 'yes' && expected === 'positive')
+	      || (answer.answer === 'no' && expected === 'negative')
 	}
       default:
 	return true
@@ -70,15 +78,17 @@ export const hardFilteredPostsForAnswer = (
 				   question:  GekanatorQuestion
 				   answer:    GekanatorAnswerValue },
 ): Post[] => {
-  if (!(questionIsFactLikeForHardFiltering (question)))
+  if (!(questionSupportsAnswerBasedHardFiltering (question)))
     return posts
 
   if (!(answer === 'yes' || answer === 'no'))
     return posts
 
   return posts.filter (post => {
-    const expected = expectedAnswerForQuestion (question, post)
-    return expected == null || expected === 'unknown' || expected === answer
+    const side = learnedSemanticSideForPost (question, post)
+    return side === 'unknown'
+	|| (answer === 'yes' && side === 'positive')
+	|| (answer === 'no' && side === 'negative')
   })
 }
 
@@ -112,11 +122,11 @@ export const recoverCandidatePosts = (
       recoveryStepCount }: { posts:                   Post[]
 			     scores:                  Map<number, number>
 			     rejectedPostIds:         Set<number>
-			     recoveredCandidatePosts: Map<number, number>
+			     recoveredCandidatePosts: Map<number, RecoveredCandidateState>
 			     eligiblePostIds:         Set<number>
 			     answerCountAtRecovery:   number
 			     recoveryStepCount:       number },
-): { recoveredCandidatePosts: Map<number, number>
+): { recoveredCandidatePosts: Map<number, RecoveredCandidateState>
      recoveryStepCount:       number } | null => {
   const recovered = new Map (recoveredCandidatePosts)
   const targetSize = nextRecoveryTargetSize (recoveryStepCount)
@@ -140,7 +150,9 @@ export const recoverCandidatePosts = (
   if (candidates.length === 0)
     return null
 
-  candidates.forEach (post => recovered.set (post.id, answerCountAtRecovery))
+  candidates.forEach (post => recovered.set (post.id, {
+      answerCountAtRecovery,
+      scoreAtRecovery: scores.get (post.id) ?? 0 }))
 
   return { recoveredCandidatePosts: recovered,
 	   recoveryStepCount: recoveryStepCount + 1 }