[Original Paper(λ Όλ¬Έ μλ³Έ)]
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1008025
[Paper Summary(λ Όλ¬Έ μμ½)]
Key Concepts(μ£Όμ κ°λ )
β Precedent(νλ‘)
- Legal decisions rely on past cases, known as precedents.(λ²μ κ²°μ μ νλ‘λΌκ³ λΆλ¦¬λ κ³Όκ±° μ¬κ±΄λ€μ μμ‘΄)
- Precedents provide guidance for judges when resolving new disputes.(νλ‘λ μλ‘μ΄ λΆμμ ν΄κ²°ν λ νμ¬λ€μκ² νλ¨ κΈ°μ€κ³Ό μ§μΉ¨μ μ 곡)
β Citation Network(μΈμ© λ€νΈμν¬)
- Cases are modeled as nodes, and citations between cases are links.(νλ‘(λλ μ¬κ±΄)λ€μ λ Έλ(μ μ )λ‘, νλ‘ κ°μ μΈμ© κ΄κ³λ λ§ν¬(μ°κ²°μ )λ‘ λͺ¨λΈλ§)
- The law can be understood as a network formed by these citation relationships.(λ²μ μ΄λ¬ν μΈμ© κ΄κ³μ μν΄ νμ±λ νλμ λ€νΈμν¬λ‘ μ΄ν΄ν μ μμ)
β Legal Centrality(λ²μ μ€μ¬μ±)
- Refers to how important or influential a case is within the legal system.(νλ‘(λλ μ¬κ±΄)κ° λ² μ²΄κ³ λ΄μμ μΌλ§λ μ€μνκ±°λ μν₯λ ₯μ΄ μλμ§λ₯Ό μλ―Έ)
- Measured by how it is embedded in the overall citation network.(μ΄λ ν΄λΉ νλ‘κ° μ 체 μΈμ© λ€νΈμν¬μμ μ΄λ»κ² μ리 μ‘κ³ μλμ§(ν¬ν¨λμ΄ μλμ§)μ λ°λΌ μΈ‘μ λ¨)
β Inward and Outward Citations(λ΄λΆ λ° μΈλΆ μΈμ©)
- Inward citations(λ΄λΆ μΈμ©): how often a case is cited by later cases(νλμ νλ‘κ° μ΄νμ νλ‘λ€μ μν΄ μΌλ§λ μμ£Ό μΈμ©λλμ§λ₯Ό λ»ν¨)
- Outward citations(μΈλΆ μΈμ©): how many prior cases a case cites(νλμ νλ‘κ° μ΄μ μ νλ‘λ€μ μΌλ§λ λ§μ΄ μΈμ©νλμ§λ₯Ό λ»ν¨)
β Centrality Measures(μ€μ¬μ± μ§ν)
Traditional(μ ν΅μ μ§ν):
- Degree centrality(μ°κ²° μ€μ¬μ±) (simple citation counts)(λ¨μ μΈμ© νμλ₯Ό μλ―Έ)
- Eigenvector centrality(κ³ μ λ²‘ν° μ€μ¬μ±) (importance of citing nodes)(ν΄λΉ νλ‘λ₯Ό μΈμ©ν λ Έλ(νλ‘)λ€μ μ€μλλ₯Ό λ°μν μ€μ¬μ±)
Proposed(μ μλ μ§ν):
- Inward relevance(λ΄λΆ κ΄λ ¨μ±) (influence)(μν₯λ ₯μ λνλ)
- Outward relevance(μΈλΆ κ΄λ ¨μ±) (foundation in law)(λ²μ κ·Όκ±° λ° ν λλ₯Ό λνλ)
1. Research Objective(μ°κ΅¬ λͺ©μ )
To develop a more accurate method for measuring the legal importance of court cases.(λ²μμ νλ‘κ° κ°λ λ²μ μ€μμ±μ λμ± μ ννκ² μΈ‘μ ν μ μλ λ°©λ²μ κ°λ°νκ³ μ ν¨. ꡬ체μ μΌλ‘ μ μλ€μ΄ λͺ©νλ‘ νλ λ°λ λ€μκ³Ό κ°μ)
Specifically, the authors aim to:
- Improve upon simple citation counts(λ¨μ μΈμ© νμ μΈ‘μ λ°©μμ κ°μ )
- Use network analysis to capture deeper structural relationships(λ€νΈμν¬ λΆμμ ν΅ν λ κΉμ ꡬ쑰μ κ΄κ³μ ν¬μ°©)
- Provide a reliable and valid measure of case importance(νλ‘ μ€μμ±μ λν μ λ’°ν μ μκ³ νλΉν μΈ‘μ μ§ν μ 곡)
2. Research Methodology(μ°κ΅¬ λ°©λ²)
1. Data Construction(λ°μ΄ν° ꡬμΆ)
- Dataset(λ°μ΄ν°μ
):
- 26,681 U.S. Supreme Court decisions (1791β2005)(26,681건μ λ―Έκ΅ μ°λ°© λλ²μ νκ²°λ¬Έ (1791λ ~2005λ ))
- Source(μΆμ²):
- Shepardβs Citations(μ °νΌλ μΈμ© μμΈ (Shepardβs Citations))
- Resul(κ²°κ³Ό)t:
- A complete citation network of Supreme Court cases(μ°λ°© λλ²μ νλ‘μ μμ ν μΈμ© λ€νΈμν¬ κ΅¬μΆ)
2. Network Modeling(λ€νΈμν¬ λͺ¨λΈλ§)
- Nodes(λ Έλ) = court cases(λ²μμ νλ‘(μ¬κ±΄))
- Directed links(λ°©ν₯μ± λ§ν¬) = citations(νλ‘ κ°μ μΈμ© κ΄κ³)
- The network reflects the structure of legal precedent(μ΄ λ€νΈμν¬λ λ²μ μ λ‘μ ꡬ쑰λ₯Ό λ°μ)
3. Existing Measures Evaluation(κΈ°μ‘΄ μΈ‘μ μ§ν νκ°)
- Degree centrality(μ°κ²° μ€μ¬μ±):
- Measures importance by citation frequency(μΈμ© λΉλλ₯Ό ν΅ν΄ μ€μλλ₯Ό μΈ‘μ )
- Limitation(νκ³): ignores citation quality(μΈμ©μ μ§(quality)μ κ³ λ €νμ§ μμ)
- Eigenvector centrality(κ³ μ λ²‘ν° μ€μ¬μ±):
- Considers importance of citing cases(μΈμ©νλ νλ‘λ€μ μ€μλκΉμ§ κ³ λ €)
- Limitation(νκ³): ignores outward citations and time bias(μΈλΆ μΈμ©(outward citations)μ κ³ λ €νμ§ μμΌλ©°, μκ°μ νΈν₯(time bias) λ¬Έμ κ° μ‘΄μ¬)
4. Proposed Method (Core Contribution)(μ μλ λ°©λ²λ‘ (ν΅μ¬ κΈ°μ¬))
The authors introduce a dual centrality model based on Kleinbergβs framework(μ μλ€μ ν΄λΌμΈλ²κ·Έ(Kleinberg)μ νλ μμν¬λ₯Ό κΈ°λ°μΌλ‘ ν μ΄μ€ μ€μ¬μ± λͺ¨λΈ(dual centrality model)μ λμ ν¨):
- Inward relevance(λ΄λΆ κ΄λ ¨μ±):
- A case is important if cited by important cases(μ€μν νλ‘λ€λ‘λΆν° μΈμ©λ νλ‘μΌμλ‘ μ€μ)
- Outward relevance(μΈλΆ κ΄λ ¨μ±):
- A case is important if it cites important cases(μ€μν νλ‘λ€μ μΈμ©ν νλ‘μΌμλ‘ μ€μ)
βΆ These two measures are mutually reinforcing(μ΄ λ μ§νλ μνΈ λ³΄μμ (μνΈ κ°νμ )):
- Influential cases are cited by well-grounded cases(μν₯λ ₯ μλ νλ‘λ λ²μ κ·Όκ±°κ° ννν νλ‘λ€μ μν΄ μΈμ©)
- Well-grounded cases cite influential cases(λ²μ κ·Όκ±°κ° ννν νλ‘λ μν₯λ ₯ μλ νλ‘λ€μ μΈμ©)
5. Validation Method(κ²μ¦ λ°©λ² λ°©λ²λ‘ )
- Statistical analysis(ν΅κ³ λΆμ):
- Negative binomial regression(μμ΄ν νκ· λΆμ (Negative binomial regression))
- Outcome variable(κ²°κ³Ό λ³μ):
- Future citations by courts(ν₯ν λ²μλ€μ μν΄ μΈμ©λλ νμ (Future citations))
βΆ Goal(λͺ©ν):
- Test whether the proposed measures predict future influence better than existing metrics(μ μλ μ§νκ° κΈ°μ‘΄ μ§νλ€λ³΄λ€ λ―Έλμ μν₯λ ₯μ λ μ μμΈ‘νλμ§ κ²μ¦νλ κ²)
3. Key Findings(μ£Όμ μ°κ΅¬ κ²°κ³Ό)
1. Network Structure(λ€νΈμν¬ κ΅¬μ‘°)
- The citation network follows a skewed distribution(μΈμ© λ€νΈμν¬λ μλ(λΉλμΉλ)κ° λμ λΆν¬λ₯Ό λ°λ¦):
- A few cases receive many citations(μμμ νλ‘κ° λλ€μμ μΈμ©μ λ°μ)
- Most cases receive few citations(λΆλΆμ νλ‘λ μΈμ©μ κ±°μ λ°μ§ λͺ»ν¨)
[κ±°λμ κ³± λ²μΉ λΆν¬μ ‘κΈ΄ 꼬리’ (λ‘±ν μΌ)]

- μΈμ© λ€νΈμν¬λ₯Ό λνλ΄λ λ°©ν₯μ± κ·Έλν μ΄λ―Έμ§
- λ€νΈμν¬ λΆμμ μ€μ¬μ± μ§νλ€μ λνλ΄λ λ€μ΄μ΄κ·Έλ¨ μ΄λ―Έμ§
- HITS μκ³ λ¦¬μ¦μ νλΈ(Hubs)μ κΆμ(Authorities)λ₯Ό λνλ΄λ λ€μ΄μ΄κ·Έλ¨ μ΄λ―Έμ§
Resource(μΆμ²): Shutterstock
2. Limitations of Simple Citation Counts(λ¨μ μΈμ© νμ μΈ‘μ μ νκ³)
- Citation counts alone do not capture(λ¨μν μΈμ©λ νμλ§μΌλ‘λ λ€μμ μμλ€μ ν¬μ°©ν μ μμ):
- The quality of citations(μΈμ©μ μ§, ν΄λΉ μΈμ©μ΄ μΌλ§λ κ°μΉ μκ³ μ€μνμ§ νλ¨νμ§ λͺ»ν¨)
- The position of a case within the network(λ€νΈμν¬ λ΄ νλ‘μ μμΉ, ν΄λΉ νλ‘κ° μ 체 μΈμ© λ€νΈμν¬ κ΅¬μ‘° μμμ μ΄λ€ μμΉλ₯Ό μ°¨μ§νκ³ μλμ§ νμ νμ§ λͺ»ν¨)
3. Superiority of the Proposed Measures(μ μλ μ§νμ μ°μμ±)
- Inward and outward relevance(λ΄λΆ κ΄λ ¨μ± λ° μΈλΆ κ΄λ ¨μ± μ§ν):
- Perform better than traditional measures(κΈ°μ‘΄μ μ ν΅μ μΈ μΈ‘μ μ§νλ€λ³΄λ€ λ μ°μν μ±λ₯μ λνλ)
- Provide stronger predictive power for future citations(ν λ€λ₯Έ νλ‘λ€μ μν΄ μΈμ©λ νλ₯ (λ―Έλμ μΈμ© νμ)μ λν΄ λ κ°λ ₯ν μμΈ‘λ ₯μ μ 곡ν¨)
4. Strong Predictive Effects(κ°λ ₯ν μμΈ‘ ν¨κ³Ό)
- Increasing inward relevance significantly raises(λ΄λΆ κ΄λ ¨μ±(inward relevance) μ§νκ° λμμ§μλ‘ λ€μμ νλ₯ μ΄ ν¬κ² μμΉ):
- The probability of a case being cited in the future(ν΄λΉ νλ‘κ° ν₯νμ μΈμ©λ νλ₯ )
- These effects are larger than those of(μ΄λ¬ν μμΈ‘ ν¨κ³Όλ λ€μ μ§νλ€μ΄ κ°μ§ ν¨κ³Όλ³΄λ€ λ νΌ):
- Citation counts(λ¨μ μΈμ© νμ)
- Other importance indicators(κΈ°ν μ€μλ μ§νλ€)
5. Key Insight(ν΅μ¬ ν΅μ°°)
Not all citations are equal(λͺ¨λ μΈμ©μ΄ λλ±ν κ°μΉλ₯Ό κ°μ§λ κ²μ μλ)
- Being cited by important cases matters more than being cited frequently(μμ£Ό μΈμ©λλ κ²λ³΄λ€ μ€μν νλ‘λ€μ μν΄ μΈμ©λλ κ²μ΄ ν¨μ¬ λ μ€μ)
4. Conclusions and Implications(κ²°λ‘ λ° μμ¬μ )
1. Main Conclusion(μ£Όμ κ²°λ‘ )
- Legal importance can be effectively measured using network-based methods(λ‘μ λ²μ μ€μμ±μ λ€νΈμν¬ κΈ°λ° λ°©λ²λ‘ μ ν΅ν΄ ν¨κ³Όμ μΌλ‘ μΈ‘μ λ μ μμ)
- The proposed measures outperform traditional approaches(λ³Έ μ°κ΅¬μμ μ μλ μ§νλ€μ κΈ°μ‘΄μ μ ν΅μ μΈ μ κ·Ό λ°©μλ€λ³΄λ€ λ μ°μν μ±λ₯μ 보μ¬μ€)
2. Theoretical Implications(μ΄λ‘ μ ν¨μ)
- Law should be understood as a network of interconnected cases(λ²μ μνΈ μ°κ²°λ νλ‘λ€μ λ€νΈμν¬λ‘ μ΄ν΄λμ΄μΌν¨)
- Legal importance is inherently relational(λ²μ μ€μμ±μ λ³Έμ§μ μΌλ‘ κ΄κ³μ (relational)μΈ μ±κ²©μ λ)
3. Methodological Contribution(λ°©λ²λ‘ μ κΈ°μ¬)
- Introduces a new, more robust measure of centrality(λμ± κ°λ ₯νκ³ μ λ’°ν μ μλ(robust) μλ‘μ΄ μ€μ¬μ± μΈ‘μ μ§νλ₯Ό λμ )
- Demonstrates the value of network analysis in legal studies(λ²ν μ°κ΅¬μμ λ€νΈμν¬ λΆμμ΄ μ§λ κ°μΉμ μ μ©μ±μ μ μ¦)
4. Practical Implications(μ€λ¬΄μ ν¨μ)
- Helps identify influential precedents(μν₯λ ₯ μλ μ λ‘λ₯Ό μ ννκ² μλ³νλ λ° λμμ μ€)
- Improves legal research and case analysis9λ²λ₯ μ°κ΅¬ λ° νλ‘ λΆμμ ν¨μ¨μ±μ ν₯μ)
- Enables prediction of future legal relevance(ν₯ν ν΄λΉ νλ‘κ° κ°μ§ λ²μ κ΄λ ¨μ±(μ€μλ)μ μμΈ‘ν μ μκ² ν¨)
5. Future Research Directions(ν₯ν μ°κ΅¬ λ°©ν₯)
- Study how precedents are overruled(νλ‘κ° μ΄λ»κ² νκΈ°(overruled)λλμ§μ λν μ°κ΅¬)
- Analyze judicial behavior in citation practices(νλ‘ μΈμ© κ΄νμ λνλλ μ¬λ²λΆμ νλ(λ²κ΄μ μ±ν₯ λ±) λΆμ)
- Compare legal networks with other complex systems(λ²λ₯ λ€νΈμν¬μ λ€λ₯Έ 볡μ‘κ³(complex systems) λ€νΈμν¬μ λΉκ΅ λΆμ)

