🧠 Boston Neuromind
νŠΉν—ˆ μΆœμ› μ§„ν–‰ 쀑 Β· 1/3Provisional Patent Application Β· 1/3

Fischer 동적 기술 μˆ˜μ€€ 적응 μ‹œμŠ€ν…œ Fischer Dynamic Skill Level Adaptation System

μ‚¬μš©μžμ˜ μ–Έμ–΄ ν‘œν˜„μœΌλ‘œλΆ€ν„° μΈμ§€Β·μ •μ„œ λ°œλ‹¬ μˆ˜μ€€μ„ 13λ‹¨κ³„λ‘œ μžλ™ μΆ”μ •ν•˜κ³ , AI μ‘λ‹΅μ˜ λ³΅μž‘λ„Β·μ–΄νœ˜Β·κ΅¬μ‘°λ₯Ό μ‹€μ‹œκ°„μœΌλ‘œ μ •ν•©μ‹œν‚€λŠ” 컴퓨터 κ΅¬ν˜„ 방법. A computer-implemented method for automatically inferring a user's cognitive-affective developmental level on a 13-tier scale from linguistic features, and dynamically calibrating an AI conversational system's response complexity, vocabulary, and structural depth in real time.
μΆœμ›μΈApplicant Boston Neuromind, LLC
발λͺ…μžInventor [발λͺ…μžλͺ…] (BCN, PhD, Ed.D.) [Inventor Name] (BCN, PhD, Ed.D.)
μƒνƒœStatus USPTO κ°€μΆœμ› μ€€λΉ„ USPTO Provisional Pending
λΆ„λ₯˜Classification G06N 20/00 / G16H 50/20 / A61B 5/377
λͺ©μ°¨Table of Contents
  1. 초둝Abstract
  2. 발λͺ… λΆ„μ•ΌField of Invention
  3. 배경 기술Background
  4. ν•΄κ²° 과제Problem Statement
  5. 발λͺ… μš”μ•½Summary of Invention
  6. 상세 μ„€λͺ…Detailed Description
  7. 도면 μ„€λͺ…Drawings
  8. 청ꡬ항Claims
  9. μ„ ν–‰ 기술 비ꡐPrior Art Comparison
  10. 산업상 이용 κ°€λŠ₯μ„±Industrial Applicability
  11. κ΄€λ ¨ λ…Όλ¬ΈReferences

01초둝Abstract

πŸ“‹ 핡심 μš”μ•½ (ν•œ 문단) πŸ“‹ One-Paragraph Summary

λ³Έ 발λͺ…은 μ‚¬μš©μžκ°€ μž…λ ₯ν•œ μžμ—°μ–΄ ν…μŠ€νŠΈ(μ±„νŒ… λ©”μ‹œμ§€, 일기, μžκ°€λ³΄κ³ )μ—μ„œ μΆ”μΆœν•œ 17개 μ–Έμ–΄ νŠΉμ§•(평균 μ–΄νœ˜ λ³΅μž‘λ„, λ¬Έμž₯ 길이 λΆ„μ‚°, 좔상 λͺ…사 λΉ„μœ¨, 메타인지 마컀 λΉˆλ„, 인과 연결사 밀도 λ“±)을 μž…λ ₯으둜 ν•˜μ—¬, Harvard λŒ€ν•™ Kurt Fischer의 동적 기술 이둠(Dynamic Skill Theory)에 κΈ°λ°˜ν•œ 13단계(Tier 0–12) λ°œλ‹¬ μˆ˜μ€€μ„ 가쀑 ν•© + μ‹œκ·Έλͺ¨μ΄λ“œ ν™œμ„±ν™” ν•¨μˆ˜λ₯Ό 톡해 μΆ”μ •ν•˜κ³ , μΆ”μ •λœ μˆ˜μ€€μ— 따라 AI 챗봇 μ‹œμŠ€ν…œμ˜ 응닡 생성 μ‹œ (a) μ–΄νœ˜ λ³΅μž‘λ„, (b) λ¬Έμž₯ ꡬ쑰 깊이, (c) λΉ„μœ μ  좔상 μ‚¬μš© λΉ„μœ¨, (d) 메타인지 μœ λ„ λΉˆλ„λ₯Ό μžλ™ μ‘°μ ˆν•˜μ—¬ μ‚¬μš©μžμ˜ λ°œλ‹¬ μ˜μ—­(Zone of Proximal Development, +1 tier)μ—μ„œμ˜ ν•™μŠ΅Β·μΉ˜λ£Œ 효과λ₯Ό κ·ΉλŒ€ν™”ν•˜λŠ” 컴퓨터 κ΅¬ν˜„ 방법 및 μ‹œμŠ€ν…œμ— κ΄€ν•œ 것이닀. The present invention relates to a computer-implemented method and system that extracts seventeen (17) linguistic features from natural language input (chat messages, journal entries, self-reports) submitted by a user β€” including mean lexical complexity, sentence length variance, abstract noun ratio, metacognitive marker frequency, and causal connective density β€” and infers a developmental level on a thirteen-tier scale (Tier 0 through Tier 12) derived from Harvard professor Kurt Fischer's Dynamic Skill Theory, via a weighted-sum-plus-sigmoid activation function; based on the inferred tier, the system dynamically calibrates an AI chatbot's response-generation parameters β€” specifically (a) vocabulary complexity, (b) syntactic structure depth, (c) figurative/abstract usage ratio, and (d) metacognitive prompt frequency β€” to maintain conversation within the user's Zone of Proximal Development (tier + 1), thereby maximizing therapeutic and educational effect.

02발λͺ… λΆ„μ•ΌField of Invention

λ³Έ 발λͺ…은 인곡지λŠ₯ 기반 λ””μ§€ν„Έ 정신건강 μ„œλΉ„μŠ€(Digital Mental Health) 및 μ μ‘ν˜• ν•™μŠ΅ μ‹œμŠ€ν…œ(Adaptive Learning Systems) 뢄야에 μ†ν•œλ‹€. λ”μš± κ΅¬μ²΄μ μœΌλ‘œλŠ”, μ‚¬μš©μžμ˜ μ–Έμ–΄ ν‘œν˜„μœΌλ‘œλΆ€ν„° λ°œλ‹¬ μˆ˜μ€€μ„ μΆ”μ •ν•˜κ³ , κ·Έ μˆ˜μ€€μ— 맞좰 AI λŒ€ν™” μ—μ΄μ „νŠΈ(LLM 기반 챗봇)의 응닡 λ§€κ°œλ³€μˆ˜λ₯Ό μžλ™ μ‘°μ •ν•˜λŠ” μ‹œμŠ€ν…œμ— κ΄€ν•œ 것이닀. The present invention pertains to the fields of AI-based Digital Mental Health Services and Adaptive Learning Systems. More specifically, it relates to a system that infers a user's developmental level from linguistic expression and automatically adjusts the response parameters of an AI conversational agent (an LLM-based chatbot) to match said level.

κ΄€λ ¨ 기술 λΆ„μ•ΌRelated Technical Fields

03배경 기술Background

3.1 Fischer 동적 기술 이둠3.1 Fischer's Dynamic Skill Theory

Kurt W. Fischer ꡐ수(Harvard Graduate School of Education)λŠ” 1980λ…„ "A theory of cognitive development: The control and construction of hierarchies of skills"(Psychological Review, 87(6), 477–531)μ—μ„œ μΈκ°„μ˜ μΈμ§€Β·μ •μ„œ λ°œλ‹¬μ΄ 13λ‹¨κ³„μ˜ μœ„κ³„μ  기술 ꡬ쑰λ₯Ό λ”°λ₯΄λ©°, 각 λ‹¨κ³„λŠ” λΉ„μ„ ν˜•μ Β·λ§₯락 의쑴적으둜 λ°œλ‹¬ν•œλ‹€λŠ” 이둠을 μ œμ‹œν–ˆλ‹€. Professor Kurt W. Fischer (Harvard Graduate School of Education) proposed in his seminal 1980 paper, "A theory of cognitive development: The control and construction of hierarchies of skills" (Psychological Review, 87(6), 477–531), that human cognitive-affective development follows a hierarchical skill structure across thirteen tiers, with each tier developing nonlinearly and context-dependently.

계측Tier 단계λͺ…Stage Name μ „ν˜•μ  μ—°λ ΉTypical Age 언어적 νŠΉμ§•Linguistic Marker
0Single Reflexes0–4moμ–Έμ–΄ 이전pre-verbal
1Reflex Mappings4–8moμ–Έμ–΄ 이전pre-verbal
2Reflex Systems8–12mo단어 λ‹¨νŽΈword fragments
3Single Sensorimotor Actions1–2y단일 단어single words
4Sensorimotor Mappings2y2-3 단어 κ²°ν•©two- to three-word combinations
5Sensorimotor Systems3–4yλ‹¨μˆœ λ¬Έμž₯simple sentences
6Single Representations4–6yꡬ체적 λͺ…사·동사concrete nouns and verbs
7Representational Mappings6–10y관계·인과 μ—°κ²°relational and causal connectives
8Representational Systems10–12y쑰건·가섀 ν‘œν˜„conditional and hypothetical expressions
9Single Abstractions12–15y좔상 λͺ…사abstract nouns
10Abstract Mappings15–19y메타인지 ν‘œν˜„metacognitive expressions
11Abstract Systems19–24y닀관점 톡합multi-perspective integration
12Single Principles24y+원리·체계 톡합principle and system synthesis

Fischer 이둠의 핡심 톡찰은 λ°œλ‹¬μ΄ "단일 μ˜μ—­μ˜ 일직선 μ§„ν–‰"이 μ•„λ‹ˆλΌ "λ§₯락별·기λŠ₯λ³„λ‘œ λ‹€λ₯Έ μˆ˜μ€€μ—μ„œ λ™μ‹œ μž‘λ™"ν•œλ‹€λŠ” 점이닀. λ”°λΌμ„œ λ™μΌν•œ μ‚¬μš©μžλ„ μ •μ„œμ  ν‘œν˜„μ—μ„œλŠ” Tier 8, 좔상 μ‚¬κ³ μ—μ„œλŠ” Tier 11에 μœ„μΉ˜ν•  수 μžˆλ‹€. The core insight of Fischer's theory is that development does not proceed as a "linear advance through a single domain"; rather, it operates simultaneously at different levels across contexts and functions. Hence the same user may operate at Tier 8 for emotional expression while operating at Tier 11 for abstract reasoning.

3.2 ZPD (κ·Όμ ‘ λ°œλ‹¬ μ˜μ—­) 원리3.2 ZPD (Zone of Proximal Development) Principle

Vygotsky(1978)의 ZPD κ°œλ…μ€ ν•™μŠ΅ νš¨κ³Όκ°€ κ·ΉλŒ€ν™”λ˜λŠ” 지점이 ν˜„μž¬ μˆ˜μ€€λ³΄λ‹€ μ•½κ°„ 높은(ν˜„μž¬ + 1) μ˜μ—­μž„μ„ μ‹œμ‚¬ν•œλ‹€. λ„ˆλ¬΄ μ‰¬μš°λ©΄ μ§€λ£¨ν•˜κ³  ν•™μŠ΅μ΄ μΌμ–΄λ‚˜μ§€ μ•ŠμœΌλ©°, λ„ˆλ¬΄ μ–΄λ €μš°λ©΄ 쒌절과 νšŒν”Όκ°€ λ°œμƒν•œλ‹€. Vygotsky's ZPD concept (1978) holds that learning is maximized when material is positioned slightly above the current level (current + 1). Material too easy yields boredom and no learning; material too difficult yields frustration and avoidance.

3.3 κΈ°μ‘΄ LLM μ±—λ΄‡μ˜ ν•œκ³„3.3 Limitations of Existing LLM Chatbots

ν˜„μž¬ μ‚¬μš© 쀑인 GPT-4, Claude, Gemini λ“±μ˜ LLM 챗봇은 λͺ¨λ‘ μ‚¬μš©μžμ˜ λ°œλ‹¬ μˆ˜μ€€μ„ λ¬΄μ‹œν•˜κ³  κ³ μ •λœ 좜λ ₯ μŠ€νƒ€μΌλ‘œ μ‘λ‹΅ν•œλ‹€. μ •μ‹ κ±΄κ°•Β·κ΅μœ‘ λ„λ©”μΈμ—μ„œ μ΄λŠ” λ‹€μŒ 문제λ₯Ό λ°œμƒμ‹œν‚¨λ‹€: Currently deployed LLM chatbots (GPT-4, Claude, Gemini, and similar systems) respond in a fixed output style that disregards the user's developmental level. In mental health and educational domains, this produces the following problems:

κΈ°μ‘΄ LLM μ‹œμŠ€ν…œμ€ "system prompt"λ₯Ό μˆ˜λ™μœΌλ‘œ μ„€μ •ν•˜μ—¬ 톀을 μ‘°μ •ν•  수 μžˆμ§€λ§Œ, μ΄λŠ” (a) μ‚¬μš©μž μž…λ ₯μœΌλ‘œλΆ€ν„° μžλ™ μΆ”μ •λ˜μ§€ μ•ŠμœΌλ©°, (b) λŒ€ν™” μ§„ν–‰ 쀑 λ™μ μœΌλ‘œ κ°±μ‹ λ˜μ§€ μ•ŠμœΌλ©°, (c) Fischer 13단계와 같은 λ°œλ‹¬ 이둠적 κ·Όκ±°κ°€ μ—†λ‹€. Existing LLM systems allow manual tone adjustment via a "system prompt," but such tuning (a) is not automatically inferred from user input, (b) is not dynamically updated as the conversation progresses, and (c) is not grounded in any developmental theoretical framework comparable to Fischer's thirteen tiers.

04ν•΄κ²° 과제Problem Statement

⚠️ λ³Έ 발λͺ…이 ν•΄κ²°ν•˜λ €λŠ” 핡심 문제⚠️ Core Problems Addressed by the Invention
  1. μžλ™ λ°œλ‹¬ μˆ˜μ€€ μΆ”μ •μ˜ λΆ€μž¬.Absence of automatic developmental-level inference. ν˜„μž¬ μ–΄λ–€ LLM μ‹œμŠ€ν…œλ„ μ‚¬μš©μžμ˜ λ°œλ‹¬ μˆ˜μ€€μ„ μžλ™ μΆ”μ •ν•˜μ—¬ 응닡 λ§€κ°œλ³€μˆ˜λ₯Ό μ‘°μ •ν•˜μ§€ μ•ŠλŠ”λ‹€. No existing LLM system automatically infers a user's developmental level and adjusts response parameters accordingly.
  2. ν‘œμ€€ν™”λœ 13단계 λ§€ν•‘μ˜ λΆ€μž¬.Absence of a standardized 13-tier mapping. μ–Έμ–΄ νŠΉμ§• β†’ λ°œλ‹¬ 단계 맀핑을 μ œκ³΅ν•˜λŠ” μ•Œκ³ λ¦¬μ¦˜μ΄ μƒμ—…μ μœΌλ‘œ μ‘΄μž¬ν•˜μ§€ μ•ŠλŠ”λ‹€. No commercially available algorithm provides a linguistic-feature-to-developmental-tier mapping.
  3. λ§₯락별 μ°¨λ“± μ μ‘μ˜ λΆ€μž¬.Absence of context-specific differential adaptation. μ‚¬μš©μžκ°€ μ •μ„œΒ·μΈμ§€Β·κ΄€κ³„ μ˜μ—­λ§ˆλ‹€ λ‹€λ₯Έ λ°œλ‹¬ μˆ˜μ€€μ— μžˆμ„ 수 μžˆμŒμ„ λ°˜μ˜ν•˜λŠ” μ‹œμŠ€ν…œμ΄ λΆ€μž¬ν•˜λ‹€. No system reflects the fact that a user may operate at different developmental tiers across emotional, cognitive, and relational contexts.
  4. μ‹€μ‹œκ°„ 동적 κ°±μ‹ μ˜ λΆ€μž¬.Absence of real-time dynamic updating. λŒ€ν™”κ°€ 진행됨에 따라 μ‚¬μš©μžμ˜ ν‘œν˜„ μˆ˜μ€€μ΄ λ³€ν™”ν•  수 μžˆμœΌλ‚˜, 이λ₯Ό μ¦‰μ‹œ λ°˜μ˜ν•˜λŠ” λ©”μ»€λ‹ˆμ¦˜μ΄ μ—†λ‹€. A user's expressive level may shift as the conversation progresses, yet no mechanism exists to immediately reflect such shifts.
  5. ZPD 기반 응닡 λ³΄μ •μ˜ λΆ€μž¬.Absence of ZPD-based response calibration. ν˜„μž¬ μˆ˜μ€€ + 1둜 응닡을 μžλ™ λ³΄μ •ν•˜μ—¬ ν•™μŠ΅Β·μΉ˜λ£Œ 효과λ₯Ό κ·ΉλŒ€ν™”ν•˜λŠ” μ‹œμŠ€ν…œμ΄ λΆ€μž¬ν•˜λ‹€. No system performs ZPD-based calibration β€” automatically setting the response one tier above the user's current level β€” to maximize learning and therapeutic effect.

05발λͺ… μš”μ•½Summary of Invention

πŸ’Ž 발λͺ…μ˜ 핡심 κ΅¬μ„±πŸ’Ž Core Components of the Invention

5.1 μ‹œμŠ€ν…œ ꡬ성 μš”μ†Œ5.1 System Components

  1. μ–Έμ–΄ νŠΉμ§• μΆ”μΆœ λͺ¨λ“ˆ (LFE)Linguistic Feature Extractor (LFE): μ‚¬μš©μž μž…λ ₯ ν…μŠ€νŠΈλ‘œλΆ€ν„° 17개 μ •λŸ‰ νŠΉμ§•μ„ μΆ”μΆœextracts seventeen (17) quantitative features from user input text
  2. Tier μΆ”λ‘  μ—”μ§„ (TIE)Tier Inference Engine (TIE): 가쀑 ν•© + μ‹œκ·Έλͺ¨μ΄λ“œλ₯Ό 톡해 0–12 단계λ₯Ό μΆ”μ •infers a 0–12 tier via weighted-sum-plus-sigmoid activation
  3. λ§₯락 뢄리 λͺ¨λ“ˆ (CSM)Context Separation Module (CSM): μ •μ„œΒ·μΈμ§€Β·κ΄€κ³„ λ„λ©”μΈλ³„λ‘œ 별도 Tier μΆ”μ •estimates separate Tiers for emotional, cognitive, and relational domains
  4. 응닡 λ§€κ°œλ³€μˆ˜ 맀퍼 (RPM)Response Parameter Mapper (RPM): Tier β†’ AI 응닡 λ§€κ°œλ³€μˆ˜(μ–΄νœ˜ λ³΅μž‘λ„, λ¬Έμž₯ 깊이, 좔상 λΉ„μœ¨, 메타인지 λΉˆλ„) λ§€ν•‘maps Tier values to AI response parameters (vocabulary complexity, syntactic depth, abstraction ratio, metacognitive prompt frequency)
  5. ZPD 보정기 (ZAC)ZPD Adjuster (ZAC): 응닡 Tier = μ‚¬μš©μž Tier + 1둜 보정calibrates response Tier as user Tier + 1
  6. 동적 κ°±μ‹  루프 (DUL)Dynamic Update Loop (DUL): λ§€ λ©”μ‹œμ§€λ§ˆλ‹€ Tier μž¬μΆ”μ • 및 응닡 λ§€κ°œλ³€μˆ˜ κ°±μ‹ re-estimates Tier and updates response parameters on each message

5.2 핡심 차별점 (Inventive Step)5.2 Inventive Steps

06상세 μ„€λͺ…Detailed Description

6.1 17개 μ–Έμ–΄ νŠΉμ§•μ˜ μ •μ˜6.1 Definition of the Seventeen Linguistic Features

각 μ‚¬μš©μž μž…λ ₯ λ©”μ‹œμ§€λ‘œλΆ€ν„° λ‹€μŒ 17개 μ •λŸ‰ νŠΉμ§•μ„ μΆ”μΆœν•œλ‹€:From each user input message, the following seventeen quantitative features are extracted:

# νŠΉμ§•λͺ…Feature μ •μ˜Definition λ°œλ‹¬ μ‹œκ·Έλ„Developmental Signal
F1MLD (Mean Lexical Difficulty)단어별 λΉˆλ„ μ—­μˆ˜ 평균average inverse word-frequencyTier ↑
F2SLV (Sentence Length Variance)λ¬Έμž₯ 길이 λΆ„μ‚°variance in sentence lengthTier ↑
F3ANR (Abstract Noun Ratio)전체 λͺ…사 쀑 좔상λͺ…사 λΉ„μœ¨ratio of abstract nouns among all nounsTier 9+
F4MMF (Metacognitive Marker Frequency)"λ‚΄ 생각", "λŠλ‚Œμƒ" λ“± λΉˆλ„frequency of markers such as "I think," "I feel like"Tier 10+
F5CCD (Causal Connective Density)"λ”°λΌμ„œ", "μ™œλƒν•˜λ©΄" 밀도density of "therefore," "because" connectivesTier 7+
F6CHD (Conditional Hypothetical Density)"λ§Œμ•½ ~라면" λΉˆλ„frequency of "if-then" constructionsTier 8+
F7SCD (Subordinate Clause Depth)μ’…μ†μ ˆ 평균 깊이average depth of subordinate clausesTier ↑
F8FUR (Figurative Usage Ratio)μ€μœ Β·λΉ„μœ  λΉ„μœ¨ratio of metaphors and similesTier 9+
F9MPI (Multi-Perspective Indicator)"ν•œνŽΈμœΌλ‘œλŠ”" λ“± 닀관점 마컀multi-perspective markers such as "on the other hand"Tier 11+
F10EVL (Emotion Vocabulary Lexicon)μ •μ„œ μ–΄νœ˜ 풍뢀도richness of emotional vocabularyTier 9+
F11TPS (Temporal Perspective Shifts)κ³Όκ±°Β·λ―Έλž˜Β·κ°€μ • μ‹œμ œ μ „ν™˜ λΉˆλ„frequency of past–future–hypothetical tense shiftsTier 8+
F12NEG (Negation Complexity)쀑첩 λΆ€μ • μ‚¬μš©use of nested negationTier 8+
F13QDF (Question Depth Factor)질문의 좔상도abstractness of questions posedTier ↑
F14SLF (Self-Reflection Frequency)자기 μ°Έμ‘° + 평가 마컀frequency of self-reference + evaluation markersTier 10+
F15SCC (System Concept Count)체계 κ°œλ… μ–΄νœ˜count of system-level concept termsTier 12
F16PRC (Principle Reference Count)원리 μ°Έμ‘° ν‘œν˜„references to principlesTier 12
F17NRD (Narrative Reasoning Depth)μ„œμ‚¬μ  인과 깊이depth of narrative causal reasoningTier ↑

6.2 Tier μΆ”λ‘  μˆ˜μ‹6.2 Tier Inference Formula

λ°œλ‹¬ 단계 Tier T(x)λŠ” λ‹€μŒ μ‹μœΌλ‘œ κ³„μ‚°ν•œλ‹€:The developmental tier T(x) is computed as follows:

μˆ˜μ‹ 1: Tier μΆ”μ • ν•¨μˆ˜Equation 1: Tier Inference Function T(x) = round( 12 Β· Οƒ( Ξ£α΅’ wα΅’ Β· Fα΅’(x) + b ) ) where: x = user input text Fα΅’ = i-th linguistic feature (i = 1..17) wα΅’ = trained weight for feature i (calibrated on Boston Neuromind clinical corpus) b = bias term Οƒ = logistic sigmoid: Οƒ(z) = 1 / (1 + e^(-z)) round(Β·) = nearest integer in {0, 1, 2, ..., 12}

6.3 λ§₯락 뢄리 μΆ”μ •6.3 Context-Separated Inference

동일 μ‚¬μš©μžμ— λŒ€ν•΄ 3개 λ„λ©”μΈλ³„λ‘œ 독립적인 Tierλ₯Ό μ‚°μΆœν•œλ‹€:Three independent Tiers are computed per user, one for each domain:

T_emotional = T(x, weights_emotional) T_cognitive = T(x, weights_cognitive) T_relational = T(x, weights_relational) β†’ Tier Vector V = (T_e, T_c, T_r)

각 도메인별 κ°€μ€‘μΉ˜(weights_emotional, weights_cognitive, weights_relational)λŠ” 17개 νŠΉμ§• 쀑 ν•΄λ‹Ή 도메인에 더 κ°•ν•œ μ‹œκ·Έλ„μ„ μ£ΌλŠ” νŠΉμ§•μ„ κ°•μ‘°ν•œλ‹€.Each domain weight set (weights_emotional, weights_cognitive, weights_relational) emphasizes the features that provide the strongest signal for that domain among the seventeen features.

6.4 응닡 λ§€κ°œλ³€μˆ˜ λ§€ν•‘6.4 Response Parameter Mapping

μΆ”μ •λœ Tier에 따라 LLM 응닡 생성 μ‹œ λ‹€μŒ 4개 λ§€κ°œλ³€μˆ˜κ°€ μžλ™ μ‘°μ ˆλœλ‹€:Based on the inferred Tier, the following four parameters are automatically adjusted at the time of LLM response generation:

λ§€κ°œλ³€μˆ˜Parameter Tier 0–4 Tier 5–8 Tier 9–12
P1: μ–΄νœ˜ λ³΅μž‘λ„Vocabulary Complexity 기초 1000 단어basic 1,000-word ν‘œμ€€ 3000 단어standard 3,000-word κ³ κΈ‰ 6000+ 단어advanced 6,000+ word
P2: λ¬Έμž₯ 깊이Syntactic Depth λ‹¨μˆœλ¬Έ μœ„μ£Όpredominantly simple 볡문 ν—ˆμš©complex clauses allowed 쀑첩 μ’…μ†μ ˆnested subordinate clauses
P3: 좔상 λΉ„μœ¨Abstraction Ratio 0–10% 10–35% 35–70%
P4: 메타인지 μœ λ„Metacognitive Prompts μ—†μŒnone 간헐적intermittent 자주frequent

6.5 ZPD 보정 (+1 Rule)6.5 ZPD Calibration (the "+1" Rule)

T_response = min( T_user + 1, 12 ) β†’ Response is generated using parameter set for T_response, not T_user, ensuring conversation stays in Zone of Proximal Development

예: μ‚¬μš©μž Tier 7 (10μ„Έ 아동 μˆ˜μ€€) β†’ 응닡 Tier 8 (쑰건·가섀 μˆ˜μ€€ ν•œ 단계 μΆ”κ°€) 으둜 μƒμ„±ν•˜μ—¬ μ‚¬μš©μžκ°€ μƒˆ μΆ”λ‘  νŒ¨ν„΄μ— λ„μ „ν•˜λ„λ‘ μœ λ„ν•œλ‹€.Example: a user at Tier 7 (10-year-old level) is met with a response generated at Tier 8 (introducing one level of conditional/hypothetical reasoning above current capacity), thereby inviting the user to engage with a new reasoning pattern.

6.6 동적 κ°±μ‹  루프6.6 Dynamic Update Loop

λ§€ μ‚¬μš©μž λ©”μ‹œμ§€ μˆ˜μ‹  μ‹œ λ‹€μŒ 절차λ₯Ό λ°˜λ³΅ν•œλ‹€:The following procedure is iterated upon receipt of each user message:

  1. ν…μŠ€νŠΈ μž…λ ₯ β†’ 17개 νŠΉμ§• μΆ”μΆœ (LFE)Text input β†’ extract seventeen features (LFE)
  2. 3개 도메인별 Tier μΆ”μ • (TIE + CSM)Estimate Tier for three domains (TIE + CSM)
  3. 졜근 5개 λ©”μ‹œμ§€ 이동 ν‰κ· μœΌλ‘œ Tier μ•ˆμ •ν™”Stabilize Tier via moving average over the most recent five messages
  4. +1 보정 적용 (ZAC)Apply +1 calibration (ZAC)
  5. 응닡 λ§€κ°œλ³€μˆ˜ μ—…λ°μ΄νŠΈ (RPM)Update response parameters (RPM)
  6. LLM에 λ§€κ°œλ³€μˆ˜ 적용된 system prompt μ£Όμž… ν›„ 응닡 생성Generate response after injecting parameter-adjusted system prompt into the LLM

07도면 μ„€λͺ…Drawings

User Input Text Message LFE Linguistic Feature Extractor (17 feat.) TIE + CSM Tier Inference 3 domains ZAC ZPD Adjuster Tier + 1 RPM Param Mapper P1–P4 LLM Response Generator parameters injected via system prompt β†’ developmentally calibrated output DUL Dynamic Update Loop 5-msg avg
도 1.FIG. 1. μ‹œμŠ€ν…œ 전체 데이터 흐름. LFE β†’ TIE β†’ CSM β†’ ZAC β†’ RPM β†’ LLM 응닡 생성. λΉ¨κ°„ 점선은 동적 κ°±μ‹  루프(DUL). Overall system data flow: LFE β†’ TIE β†’ CSM β†’ ZAC β†’ RPM β†’ LLM response generation. Red dashed line indicates the Dynamic Update Loop (DUL).
Fischer 13-Tier Developmental Hierarchy Tier 0–4 Sensorimotor (0–2y) Tier 5–8 Representational (2–12y) Tier 9–11 Abstract (12–24y) Tier 12 Principles (24y+) ZPD: Response = User Tier + 1 User detected at Tier 7 Response calibrated to Tier 8 Conversation maintained in Zone of Proximal Development
도 2.FIG. 2. Fischer 13단계 μœ„κ³„μ™€ ZPD +1 보정 원리. Fischer's 13-tier hierarchy and the ZPD +1 calibration principle.

08청ꡬ항Claims

청ꡬ항 1 (독립항)Claim 1 (Independent)
μ‚¬μš©μžμ˜ μžμ—°μ–΄ μž…λ ₯에 μ‘λ‹΅ν•˜λŠ” 인곡지λŠ₯ λŒ€ν™” μ‹œμŠ€ν…œμ— μ˜ν•΄ μˆ˜ν–‰λ˜λŠ”, 컴퓨터 κ΅¬ν˜„ μ μ‘ν˜• 응닡 생성 λ°©λ²•μœΌλ‘œμ„œ:
(a) μ‚¬μš©μž μž…λ ₯ ν…μŠ€νŠΈλ‘œλΆ€ν„° 평균 μ–΄νœ˜ λ³΅μž‘λ„, λ¬Έμž₯ 길이 λΆ„μ‚°, 좔상 λͺ…사 λΉ„μœ¨, 메타인지 마컀 λΉˆλ„ 및 인과 연결사 밀도λ₯Ό ν¬ν•¨ν•˜λŠ” 볡수의 μ •λŸ‰μ  μ–Έμ–΄ νŠΉμ§•μ„ μΆ”μΆœν•˜λŠ” 단계;
(b) μΆ”μΆœλœ μ–Έμ–΄ νŠΉμ§•μ˜ 가쀑 합에 λΉ„μ„ ν˜• ν™œμ„±ν™” ν•¨μˆ˜λ₯Ό μ μš©ν•˜μ—¬, Fischer 동적 기술 μ΄λ‘ μ—μ„œ μ •μ˜λœ 0λΆ€ν„° 12κΉŒμ§€μ˜ 13개 λ°œλ‹¬ 단계 쀑 μ‚¬μš©μžμ˜ ν˜„μž¬ λ°œλ‹¬ 단계 좔정값을 μ‚°μΆœν•˜λŠ” 단계;
(c) μΆ”μ •λœ λ°œλ‹¬ 단계에 1을 κ°€μ‚°ν•˜μ—¬ 응닡 단계λ₯Ό κ²°μ •ν•˜λ˜, μ΅œλŒ“κ°’μ„ 12둜 μ œν•œν•˜λŠ” 단계;
(d) κ²°μ •λœ 응닡 단계에 따라 인곡지λŠ₯ 응닡 μƒμ„±κΈ°μ˜ (i) μ–΄νœ˜ λ³΅μž‘λ„, (ii) λ¬Έμž₯ ꡬ쑰 깊이, (iii) 좔상 ν‘œν˜„ λΉ„μœ¨ 및 (iv) 메타인지 μœ λ„ λΉˆλ„λ₯Ό ν¬ν•¨ν•˜λŠ” 응닡 λ§€κ°œλ³€μˆ˜ 집합을 μžλ™ μ‘°μ •ν•˜λŠ” 단계; 및
(e) μ‘°μ •λœ 응닡 λ§€κ°œλ³€μˆ˜λ₯Ό μ μš©ν•˜μ—¬ 응닡 ν…μŠ€νŠΈλ₯Ό μƒμ„±ν•˜λŠ” 단계;
λ₯Ό ν¬ν•¨ν•˜λŠ” 것을 νŠΉμ§•μœΌλ‘œ ν•˜λŠ” 방법.
A computer-implemented method, performed by an artificial intelligence conversational system responding to natural language input from a user, the method comprising:
(a) extracting, from a user input text, a plurality of quantitative linguistic features including mean lexical complexity, sentence length variance, abstract noun ratio, metacognitive marker frequency, and causal connective density;
(b) computing a weighted sum of the extracted linguistic features and applying a nonlinear activation function thereto, thereby yielding an estimated current developmental tier of the user, said tier being one of thirteen tiers (Tier 0 through Tier 12) defined in Fischer's Dynamic Skill Theory;
(c) determining a response tier by adding one (1) to the estimated current developmental tier, subject to a maximum of twelve (12);
(d) automatically adjusting a set of response parameters of an artificial intelligence response generator according to the determined response tier, said set comprising (i) vocabulary complexity, (ii) syntactic structure depth, (iii) abstract-expression ratio, and (iv) metacognitive prompt frequency; and
(e) generating a response text by applying the adjusted response parameters.
청ꡬ항 2 (쒅속항)Claim 2 (Dependent)
청ꡬ항 1에 μžˆμ–΄μ„œ, 단계 (a)λŠ” 17개의 μ •λŸ‰μ  μ–Έμ–΄ νŠΉμ§•μ„ μΆ”μΆœν•˜κ³ , 상기 17개 νŠΉμ§•μ€ 평균 μ–΄νœ˜ λ‚œμ΄λ„, λ¬Έμž₯ 길이 λΆ„μ‚°, 좔상 λͺ…사 λΉ„μœ¨, 메타인지 마컀 λΉˆλ„, 인과 연결사 밀도, 쑰건/κ°€μ„€ ν‘œν˜„ 밀도, μ’…μ†μ ˆ 깊이, λΉ„μœ  μ‚¬μš© λΉ„μœ¨, 닀관점 μ§€ν‘œ, μ •μ„œ μ–΄νœ˜ 풍뢀도, μ‹œμ œ μ „ν™˜ λΉˆλ„, λΆ€μ • λ³΅μž‘λ„, 질문 깊이, μžκΈ°μ„±μ°° λΉˆλ„, 체계 κ°œλ… 수, 원리 μ°Έμ‘° 수 및 μ„œμ‚¬μ  μΆ”λ‘  깊이λ₯Ό ν¬ν•¨ν•˜λŠ” 것을 νŠΉμ§•μœΌλ‘œ ν•˜λŠ” 방법. The method of Claim 1, wherein step (a) extracts seventeen (17) quantitative linguistic features, said seventeen features comprising: mean lexical difficulty; sentence length variance; abstract noun ratio; metacognitive marker frequency; causal connective density; conditional/hypothetical density; subordinate clause depth; figurative usage ratio; multi-perspective indicator; emotion vocabulary richness; temporal perspective shifts; negation complexity; question depth factor; self-reflection frequency; system concept count; principle reference count; and narrative reasoning depth.
청ꡬ항 3 (쒅속항)Claim 3 (Dependent)
청ꡬ항 1에 μžˆμ–΄μ„œ, 단계 (b)λŠ” μ •μ„œμ  도메인, 인지적 도메인 및 관계적 도메인에 λŒ€ν•΄ 각각 λ³„κ°œμ˜ κ°€μ€‘μΉ˜ 집합을 μ μš©ν•˜μ—¬ 3개의 독립적인 λ°œλ‹¬ 단계 좔정값을 μ‚°μΆœν•˜κ³ , 단계 (d)의 응닡 λ§€κ°œλ³€μˆ˜ 쑰정은 도메인별 좔정값에 κΈ°λ°˜ν•˜μ—¬ μˆ˜ν–‰λ˜λŠ” 것을 νŠΉμ§•μœΌλ‘œ ν•˜λŠ” 방법. The method of Claim 1, wherein step (b) computes three independent developmental-tier estimates by applying distinct weight sets for an emotional domain, a cognitive domain, and a relational domain, respectively, and wherein the response-parameter adjustment of step (d) is performed based on the domain-specific estimates.
청ꡬ항 4 (쒅속항)Claim 4 (Dependent)
청ꡬ항 1에 μžˆμ–΄μ„œ, μ‚¬μš©μžμ™€μ˜ λŒ€ν™”κ°€ μ§„ν–‰λ˜λŠ” λ™μ•ˆ λ§€ μ‚¬μš©μž μž…λ ₯ λ©”μ‹œμ§€λ§ˆλ‹€ 단계 (a) λ‚΄μ§€ (e)λ₯Ό λ°˜λ³΅ν•˜κ³ , 단계 (b)의 λ°œλ‹¬ 단계 좔정값을 졜근 N개(Nβ‰₯2) λ©”μ‹œμ§€μ— λŒ€ν•œ 이동 ν‰κ· μœΌλ‘œ μ•ˆμ •ν™”ν•˜λŠ” 것을 νŠΉμ§•μœΌλ‘œ ν•˜λŠ” 방법. The method of Claim 1, further comprising iterating steps (a) through (e) for each user input message during the course of a conversation, and stabilizing the developmental-tier estimate of step (b) by computing a moving average over the most recent N messages, where N is greater than or equal to two.
청ꡬ항 5 (쒅속항)Claim 5 (Dependent)
청ꡬ항 1에 μžˆμ–΄μ„œ, 단계 (b)의 λΉ„μ„ ν˜• ν™œμ„±ν™” ν•¨μˆ˜λŠ” λ‘œμ§€μŠ€ν‹± μ‹œκ·Έλͺ¨μ΄λ“œ ν•¨μˆ˜μΈ Οƒ(z) = 1 / (1 + e^(-z)) 이고, λ°œλ‹¬ 단계 좔정값은 12 Β· Οƒ(가쀑합 + 편ν–₯)을 κ°€μž₯ κ°€κΉŒμš΄ μ •μˆ˜λ‘œ λ°˜μ˜¬λ¦Όν•˜μ—¬ κ²°μ •λ˜λŠ” 것을 νŠΉμ§•μœΌλ‘œ ν•˜λŠ” 방법. The method of Claim 1, wherein the nonlinear activation function of step (b) is the logistic sigmoid function Οƒ(z) = 1 / (1 + e^(-z)), and the developmental-tier estimate is determined by rounding 12Β·Οƒ(weighted sum + bias) to the nearest integer.
청ꡬ항 6 (쒅속항)Claim 6 (Dependent)
청ꡬ항 1에 μžˆμ–΄μ„œ, 단계 (e)의 응닡 ν…μŠ€νŠΈ 생성은 λŒ€ν˜• μ–Έμ–΄ λͺ¨λΈ(LLM)에 μ‹œμŠ€ν…œ ν”„λ‘¬ν”„νŠΈλ₯Ό μ£Όμž…ν•¨μœΌλ‘œμ¨ μˆ˜ν–‰λ˜λ©°, 상기 μ‹œμŠ€ν…œ ν”„λ‘¬ν”„νŠΈλŠ” 단계 (d)μ—μ„œ κ²°μ •λœ 응닡 λ§€κ°œλ³€μˆ˜μ— 따라 λ™μ μœΌλ‘œ κ΅¬μ„±λ˜λŠ” 것을 νŠΉμ§•μœΌλ‘œ ν•˜λŠ” 방법. The method of Claim 1, wherein the response text generation of step (e) is performed by injecting a system prompt into a Large Language Model (LLM), said system prompt being dynamically composed according to the response parameters determined in step (d).
청ꡬ항 7 (독립항 β€” μ‹œμŠ€ν…œ)Claim 7 (Independent β€” System)
청ꡬ항 1 λ‚΄μ§€ 6 쀑 μ–΄λŠ ν•œ ν•­μ˜ 방법을 μˆ˜ν–‰ν•˜κΈ° μœ„ν•œ, 적어도 ν•˜λ‚˜μ˜ ν”„λ‘œμ„Έμ„œμ™€ 상기 ν”„λ‘œμ„Έμ„œμ— μ˜ν•΄ μ‹€ν–‰λ˜λŠ” λͺ…λ Ήμ–΄λ₯Ό μ €μž₯ν•˜λŠ” λΉ„μΌμ‹œμ  컴퓨터 νŒλ… κ°€λŠ₯ μ €μž₯ 맀체λ₯Ό ν¬ν•¨ν•˜λŠ” μ μ‘ν˜• 인곡지λŠ₯ λŒ€ν™” μ‹œμŠ€ν…œ. An adaptive artificial intelligence conversational system for performing the method of any one of Claims 1 through 6, the system comprising at least one processor and a non-transitory computer-readable storage medium storing instructions executable by the processor.
청ꡬ항 8 (독립항 β€” 맀체)Claim 8 (Independent β€” Medium)
컴퓨터에 μ˜ν•΄ 싀행될 λ•Œ 청ꡬ항 1 λ‚΄μ§€ 6 쀑 μ–΄λŠ ν•œ ν•­μ˜ 방법을 μˆ˜ν–‰ν•˜λ„λ‘ ν•˜λŠ” λͺ…λ Ήμ–΄λ₯Ό μ €μž₯ν•˜λŠ” λΉ„μΌμ‹œμ  컴퓨터 νŒλ… κ°€λŠ₯ μ €μž₯ 맀체. A non-transitory computer-readable storage medium storing instructions which, when executed by a computer, cause the computer to perform the method of any one of Claims 1 through 6.

09μ„ ν–‰ 기술 비ꡐPrior Art Comparison

μ„ ν–‰ 기술Prior Art μ ‘κ·Ό 방식Approach ν•œκ³„Limitation λ³Έ 발λͺ…κ³Όμ˜ 차이Distinction from Invention
OpenAI GPT-4 (2023)
System prompts
μˆ˜λ™ 톀 μ‘°μ •manual tone adjustment μžλ™ μΆ”μ • μ—†μŒ, λ°œλ‹¬ 이둠 λΆ€μž¬no auto-inference; no developmental grounding 17 νŠΉμ§• μžλ™ μΆ”μΆœ + 13단계 λ§€ν•‘17-feature auto-extraction + 13-tier mapping
Khan Academy Khanmigo (2023) 학년별 사전 μ„€μ •grade-level presets 정적, μ‚¬μš©μž μž…λ ₯ 기반 μ‘°μ • μ—†μŒstatic; no input-based adjustment μ‹€μ‹œκ°„ 동적 κ°±μ‹  + ZPD +1real-time dynamic updating + ZPD +1
Replika / Character.AI κ³ μ • 페λ₯΄μ†Œλ‚˜fixed personas μ‚¬μš©μž λ°œλ‹¬ μˆ˜μ€€ 무관decoupled from user developmental level λ§₯락별 도메인 뢄리 μΆ”μ •domain-separated context-specific inference
Woebot / Wysa
(CBT 챗봇)
κ³ μ • 슀크립트fixed scripts μ–Έμ–΄ 적응 μ—†μŒno linguistic adaptation κ°œλ³„ λ©”μ‹œμ§€λ§ˆλ‹€ λ§€κ°œλ³€μˆ˜ κ°±μ‹ per-message parameter updating
Lexile / Flesch-Kincaid
(가독성 점수)
ν…μŠ€νŠΈ λ‚œμ΄λ„ μΈ‘μ •text difficulty scoring 좜λ ₯ μΈ‘μ •λ§Œ κ°€λŠ₯, μž…λ ₯β†’μΆœλ ₯ λ§€ν•‘ μ—†μŒoutput measurement only; no input-to-output mapping μ‚¬μš©μž μž…λ ₯ β†’ AI 응닡 λ§€κ°œλ³€μˆ˜ 직접 λ§€ν•‘direct mapping from user input to AI response parameters
🎯 발λͺ…μ˜ 진보성 (Inventive Step)🎯 Inventive Step

λ³Έ 발λͺ…은 (1) Fischer 13단계 λ°œλ‹¬ 이둠을 LLM 응닡 λ§€κ°œλ³€μˆ˜μ™€ 직접 λ§€ν•‘ν•œ 졜초의 μ‹œμŠ€ν…œμ΄λ©°, (2) μ‚¬μš©μž μž…λ ₯μœΌλ‘œλΆ€ν„° μžλ™ μΆ”μ •λœ λ°œλ‹¬ 단계에 따라 응닡을 ZPD(+1)둜 λ³΄μ •ν•˜κ³ , (3) μ •μ„œΒ·μΈμ§€Β·κ΄€κ³„ 도메인을 λΆ„λ¦¬ν•˜μ—¬ 닀차원 적응을 μˆ˜ν–‰ν•œλ‹€. 이 μ„Έ κ°€μ§€ 결합은 μ„ ν–‰ κΈ°μˆ μ— μ‘΄μž¬ν•˜μ§€ μ•ŠμœΌλ©°, 발λͺ…μ˜ λΉ„μžλͺ…μ„±(non-obviousness)을 κ΅¬μ„±ν•œλ‹€. The present invention is (1) the first system to directly map Fischer's 13-tier developmental theory to LLM response parameters; (2) it calibrates responses to ZPD (+1) based on developmental tiers automatically inferred from user input; and (3) it performs multi-dimensional adaptation by separating the emotional, cognitive, and relational domains. The combination of these three elements does not exist in the prior art and establishes the non-obviousness of the invention.

10산업상 이용 κ°€λŠ₯μ„±Industrial Applicability

10.1 적용 μ‹œμž₯10.1 Target Markets

10.2 κ΅¬ν˜„ ν˜•νƒœ10.2 Implementations

10.3 이둠적 기반10.3 Theoretical Basis

λ³Έ 발λͺ…은 발λͺ…μž(Boston Neuromind μ„€λ¦½μž)κ°€ Harvard Graduate School of Education의 Visiting Scholarλ‘œμ„œ ζ•… Kurt Fischer ꡐ수의 직접 μ§€λ„ν•˜μ— 동적 기술 이둠을 ν•™μŠ΅ν•˜κ³ , λ‡ŒνŒŒ ERP 기반 ν•™μŠ΅ μ΅œμ ν™” 연ꡬ에 μ°Έμ—¬ν•œ κ²½ν—˜μ„ ν† λŒ€λ‘œ κ΅¬μƒλ˜μ—ˆλ‹€. 이둠적 기반의 κΉŠμ΄μ™€ μž„μƒμ  κ²€μ¦μ˜ 결합이 λ³Έ 발λͺ…μ˜ 독창성을 λ’·λ°›μΉ¨ν•œλ‹€. The invention was conceived by the inventor β€” founder of Boston Neuromind β€” based on direct mentorship with the late Professor Kurt Fischer at the Harvard Graduate School of Education as Visiting Scholar, where the inventor studied Dynamic Skill Theory firsthand and participated in EEG/ERP-based learning-optimization research. The depth of theoretical grounding combined with clinical validation underpins the originality of this invention.

11κ΄€λ ¨ λ…Όλ¬ΈReferences

λ³Έ 발λͺ…μ˜ μ΄λ‘ μ Β·μž„μƒμ  κ·Όκ±°κ°€ λ˜λŠ” 핡심 λ…Όλ¬Έ 및 자료. ν΄λ¦­ν•˜λ©΄ μ™ΈλΆ€ 좜처둜 μ΄λ™ν•©λ‹ˆλ‹€. Key papers and resources providing the theoretical and clinical basis for this invention. Click links to access external sources.

A. Fischer 동적 기술 이둠A. Fischer Dynamic Skill Theory
B. Vygotsky ZPD 및 λ°œλ‹¬ ν•™μŠ΅ 이둠B. Vygotsky ZPD and Developmental Learning
C. μ–Έμ–΄ λ³΅μž‘λ„ 및 가독성 μΈ‘μ •C. Linguistic Complexity and Readability
D. LLM 및 λŒ€ν™”ν˜• AID. LLMs and Conversational AI
E. μ μ‘ν˜• ν•™μŠ΅ μ‹œμŠ€ν…œE. Adaptive Learning Systems
F. λ””μ§€ν„Έ 정신건강F. Digital Mental Health