Methodology

WHAT IS THE KNOWLEDGE GRAPH

At the core of THE GRAPHS is a knowledge graph — a network database that maps relationships between players, coaches, clubs, agents, and events across La Liga and the Premier League.

Unlike traditional databases that store information in rows and columns, a graph database (powered by Neo4j) stores entities as nodes and relationships as edges. Each edge carries a weight representing the strength of the connection, updated in real time as new data flows in.

This structure allows us to ask questions no traditional database can answer efficiently: “Which players share an agent who also represents a coach at a rival club?” or “How many degrees of separation exist between a transfer target and their rumored destination?”

HOW SIGNALS ARE DETECTED

A Graph Signal is a non-obvious connection or pattern detected by our system. Signals are generated when:

• Edge weight changes — a relationship between two entities strengthens or weakens beyond a threshold (e.g., increased social media mentions linking a player to a new club)
• New edges form — a previously unconnected pair of entities becomes linked through a shared event, agent, or data point
• Cluster patterns emerge — multiple related edges shift simultaneously, suggesting a coordinated event (e.g., a managerial change affecting multiple player connections)
• Temporal anomalies — a connection that historically strengthens at a specific time (e.g., transfer windows) activates outside that window

Each signal is classified by type: transfer, tactical, rivalry, form, injury, governance, or match-related. The system processes signals continuously, with our AI orchestrator (NEXUS) evaluating the graph state three times daily.

CONFIDENCE SCORES

Every article published by THE GRAPHS carries a confidence score from 0% to 100%. This score is not editorial opinion — it is a calculated measure based on:

• Signal strength — the magnitude of the graph change that triggered the story
• Source corroboration — how many independent data sources support the claim (minimum 2 required)
• Historical accuracy — how reliably similar signals have predicted real outcomes in the past
• Recency — how fresh the underlying data is

Articles are only published when the confidence score reaches 75% or higher. We display this score transparently so readers can calibrate their own judgment.

DATA SOURCES

THE GRAPHS aggregates data from multiple public sources to build and maintain the knowledge graph:

• Football data APIs — match results, standings, fixtures, player statistics, and team data from football-data.org and StatsBomb open data
• Social media — public posts from X/Twitter mentioning tracked entities, processed for sentiment and entity extraction
• Press conferences — YouTube transcripts from official club channels and press briefings
• Match events — detailed event data including goals, assists, cards, substitutions, and tactical formations

We do not use private, leaked, or illegally obtained information. All data sources are publicly available.

EDITORIAL STANDARDS

Every article passes six mandatory validation checks before publication:

1. Sensationalism check — the headline must not overstate the signal
2. Source grounding — every claim must be backed by source data, with a minimum of 2 corroborating sources
3. Confidence threshold — signal strength must be 75% or higher
4. Defamation risk — no personal attacks or unverified accusations
5. Inferential language — claims are framed as such (“signals suggest”, “graph indicates”); no absolutes without official confirmation
6. Balance — no bias without data justification

THE GRAPHS is AI-generated journalism with human editorial oversight. Every article includes a “Why This Story?” panel showing the signal type, confidence score, graph edge, and source data that triggered the story.

AI TRANSPARENCY BADGES

Each article displays verification badges that indicate editorial rigor and AI transparency. These badges provide immediate visual confirmation of our standards:

This transparency helps readers understand the editorial process behind each article and builds trust through clarity about AI involvement and source verification.

THE “WHY THIS STORY?” PANEL

Every article on THE GRAPHS includes an expandable panel that shows:

• The signal type that triggered the article
• The confidence score and what it means
• The graph edge — which entities are connected and how
• A mini graph visualization showing the entity neighborhood

This level of transparency is unprecedented in sports media. We believe readers deserve to know not just what the story is, but why and how it was detected.

LIMITATIONS

We are transparent about what our system cannot do:

• Graph signals are probabilistic, not certain. A high confidence score means the data strongly supports a conclusion, not that it is guaranteed.
• Our system detects connections and patterns. It does not have access to private negotiations, closed-door meetings, or confidential information.
• AI-generated text can occasionally produce awkward phrasing or miss nuance that a human journalist would catch. We continuously improve our validation pipeline.
• Historical data quality varies by league and era. Coverage is strongest for La Liga and the Premier League from 2020 onward.