Precedential Weight

The single most deterministic factor. A decision from a higher court within the same jurisdictional hierarchy is binding precedent on lower courts. Computationally, this is modeled as a jurisdictional filtering constraint where the court level attribute defines mandatory edges in the authority graph.

• Supreme Court: Maximum weight, universal binding authority within the jurisdiction.
• Appellate Court: Binding on district courts within its circuit; persuasive elsewhere.
• Trial Court: Generally zero binding weight; may carry persuasive weight for similarly situated courts.

A decision only binds courts within its sovereign territory. Persuasive authority arises when a court looks to a well-reasoned decision from a different jurisdiction. Algorithms apply jurisdictional filtering to ensure authority scores reflect only legally relevant precedent, preventing cross-sovereign contamination in binding analysis.

• Binding Scope: Limited to the issuing court's geographic or subject-matter jurisdiction.
• Persuasive Scope: Weighted by the reputation of the issuing court and the similarity of the legal question.

A case's weight is dynamic and history-dependent. Shepardizing and treatment type classification algorithms analyze how later courts have treated the decision. Positive treatment reinforces weight; negative treatment diminishes or nullifies it.

• Negative Treatment Signals: 'Overruled,' 'Abrogated,' or 'Questioned' statuses trigger a catastrophic drop in authority score.
• Positive Treatment: 'Followed,' 'Affirmed,' or 'Applied' signals increase centrality and influence.
• Distinguishing: A neutral-to-negative signal where a court declines to apply precedent due to factual differences, modeled as a weighted edge attribute.

Graph-theoretic metrics quantify structural importance. Authority propagation algorithms like PageRank variants distribute influence across the network. A case cited by many other highly authoritative cases receives a high score.

• Betweenness Centrality: Identifies cases that act as critical bridges between distinct doctrinal clusters or circuits.
• In-Degree Centrality: Raw count of citing cases, a baseline measure of visibility.
• Eigenvector Centrality: Measures influence based on the authority of citing nodes, not just volume.

Legal authority is not timeless. Temporal citation analysis models how influence decays or grows. A seminal case exhibits sustained high citation velocity over decades, while an aging precedent may lose relevance.

• Citation Velocity: The rate of new citations over time. A sudden spike may indicate renewed relevance or controversy.
• Precedent Aging: A decay function applied to older decisions that have not been recently cited, reducing their weight in predictive models.
• Citation Cascade: A seminal decision triggers a chain reaction of citations that propagates through the legal system over time.

Not all citations are equal. Citation intent classification determines the rhetorical purpose, while citation sentiment analyzes polarity. A citation used for critical disagreement has a different weight impact than one used for legal support.

• Supportive Intent: 'Followed,' 'Applied,' 'Relied Upon' — reinforces weight.
• Neutral Intent: 'Discussed,' 'Cited,' 'Explained' — minimal weight impact.
• Negative Intent: 'Criticized,' 'Questioned,' 'Overruled' — actively diminishes weight.

The computational representation of the doctrine requiring courts to follow precedent. This involves encoding jurisdictional hierarchies and binding constraints into graph structures, enabling AI systems to predict which prior decisions a court is obligated to apply. The model must distinguish between vertical stare decisis (higher courts binding lower courts) and horizontal stare decisis (a court's relationship to its own prior decisions).

A graph algorithm that iteratively distributes precedential influence scores across a citation network. Often using PageRank variants like Topic-Sensitive PageRank, it weights edges by treatment type and citation sentiment to identify the most legally significant nodes. This process accounts for the principle that a citation from a highly authoritative source confers more weight than one from a lower court.

An NLP task that automatically categorizes how a citing case legally treats a cited authority. Key labels include:

• Overruled: Prior holding explicitly invalidated
• Distinguished: Precedent not applied due to factual differences
• Followed: Court adheres to the prior ruling
• Criticized: Court questions the reasoning without overruling This classification directly modulates the edge weight in a citation graph.

A fundamental distinction in jurisdictional filtering. Binding precedent is a decision from a higher court within the same jurisdiction that a lower court must follow. Persuasive authority comes from outside the binding hierarchy and may be considered but not required. Computational models enforce this by applying hard constraints during graph traversal, ensuring authority scores reflect only legally relevant influence.

The algorithmic identification of landmark decisions that serve as the origin points for major legal doctrines. These nodes are characterized by high out-degree centrality, sustained citation velocity over decades, and a position at the root of large citation cascades. Detecting them is critical for understanding the foundational structure of a legal domain.

A composite metric that quantifies the total jurisprudential impact of a single decision. It aggregates multiple signals:

• Raw citation count
• Weighted authority score from propagation algorithms
• Treatment sentiment polarity
• Temporal decay factors This provides a multi-dimensional view of a case's true precedential weight beyond simple citation frequency.