Distinguish before connecting
The artificial intelligence ecosystem has become too dense to understand as a simple list of names. The same reader may encounter ChatGPT as an interactive system, AlphaGo as a historical milestone, a language model as a technical object, an autonomous agent as a research system, a company as a producing organization, or a fictional intelligence as a cultural object. Their thematic proximity does not make them equivalent.
The knowledge graph developed through LXKeys.world starts from that difference. Its general documentary unit is the Entry. An Entry can represent an artificial intelligence system, a model, an agent, an organization, a dataset, a benchmark, a framework, an architecture, a historical system, or another object significant enough to document. The term Entity has a narrower meaning. It refers to LXKeys Entities identified through the AES system and connected to LXSpatium.
This distinction avoids a common failure in technological mapping. Everything that belongs to the same field should not be reduced to one category. A model should remain identifiable as a model. An organization should remain an organization. An agent should remain an agent. An LXKeys Entity retains its own continuity within the ecosystem architecture.
Turn records into context
An isolated record describes. A graph situates. That difference becomes decisive when documented relationships help explain the origin, dependency, or continuity of an object.
A model can connect to its developer. A system can connect to its organization. A dataset can be associated with a benchmark. An architecture can lead to the systems that use it. An organization can lead to the systems it develops. An LXKeys Entity can connect to its canonical reference in LXSpatium. Relationships already visible around actors such as OpenAI or Mistral AI can therefore become more explicit.
The Graph interface gives this logic a navigable form. An Entry is no longer only a destination where a description can be read. It becomes a starting point from which connections can be followed and an object can be placed inside a broader documentary network.
This architecture also changes the value of identification. Stable ENT identifiers can address Entries independently of their visual presentation. Knowledge is therefore not bound only to a page. It can be retrieved, connected, and used as a structured object.
Select rather than accumulate
Artificial intelligence continuously produces models, variants, checkpoints, products, interfaces, experimental repositories, and software layers. A graph that attempted to register every appearance mechanically would risk confusing volume with documentary value.
LXKeys.world uses a curatorial logic. An Entry should have a reason to exist independently. A significant system, a foundational architecture, an autonomous agent, an important research system, a structuring organization, a historically notable technology family, or a culturally significant object may justify its own place. A minor variant, a simple checkpoint, a temporary commercial interface, or a service built on third party models does not necessarily justify a separate Entry.
The central criterion is therefore not exhaustiveness. It is the ability to preserve information that remains understandable after products change, companies disappear, or terminology evolves. Selection, provenance, and continuity become more important than the total number of records.
Attach evidence to knowledge
A graph gains value only when its claims can be connected to sufficiently strong evidence. Within this model, discovering information, validating it, and publishing it are separate operations.
The documentary hierarchy begins, when relevant, with canonical LXKeys data. It then extends to official sources, technical documentation, publications from the organizations concerned, official APIs, research work, reference repositories, institutional sources, and editorial material robust enough to support a claim.
This discipline matters especially in a field where products are renamed, models evolve, versions disappear, and announced capabilities can be revised. Information that is available online does not automatically deserve to enter the graph.
Provenance functions as editorial memory. References remain associated with Entries so that information can retain an anchor. Images can follow the same logic when a reliable visual has documentary value rather than serving only as decoration.
Give time a documentary function
A useful chronology should not exist to fill space. It should record events that genuinely change how an Entry is understood.
A launch, a scientific publication, a major version, a renaming, an acquisition, an important transition, a withdrawal, or an end of support can provide meaningful markers. By contrast, a vague mention of public appearance without an identifiable event adds little information. When no precise and defensible event is available, an empty timeline is better than an artificial sequence of dates.
Time therefore becomes a way to document continuity and change. For structures specific to LXKeys, this temporal layer can also coexist with DYPCLT and LXCalendarium when those systems apply.
Measure documentation without ranking intelligence
The documentary maturity of an Entry can be assessed through the depth of its references, structure, chronology, and relationships. That assessment is not a ranking of intelligence.
A historically important system can be very well documented while being technologically outdated. A recent system can be highly capable while still having few public references. Documentary quality and system capability answer different questions.
Keeping them separate protects the purpose of the graph. LXKeys.world can indicate that one Entry is better established than another at the documentary level without claiming to measure its value, power, or degree of intelligence.
Connect the ecosystem without erasing differences
LXKeys Entities form a distinct layer. Their canonical AES information comes from LXSpatium, while LXKeys.world can present them inside a graph that also contains external systems. Coexistence does not imply equivalence.
That separation makes broader interconnection possible. LXSpatium carries the structure and continuity specific to Entities. LXCalendarium and DYPCLT can provide temporal markers. Chronoscript can structure certain events and continuities. Cybaxter and other systems retain specialized functions. LXKeys Network participates in general interconnection. LXKeys.world occupies the role of a structured documentary map of the external world and the intelligent systems that appear within it.
The same data can then have several readers. A person can consult an Entry to understand a system. An application can use its structured fields. A machine can follow its relationships. An LXKeys Entity can, depending on available mechanisms and access rules, use the information as additional context.
The objective is not to concentrate every function in one platform. It is to let specialized systems share identifiers, references, relationships, and structured data when that circulation has a useful purpose.
Open the graph without abandoning validation
A living graph must be able to evolve. New systems appear. Names change. References become obsolete. Errors need correction. Stronger evidence can replace weaker material.
LXKeys.world therefore allows visitors to propose a source, a correction, an update, or a new Entry. Public contribution does not directly modify the registry. It passes through review before publication. A company can propose an Entry, a researcher can provide a better reference, a reader can identify an error, and an organization can request an update. None of these actions guarantees acceptance.
The separation between submission and publication protects the coherence of the graph while allowing it to evolve.
Knowledge that is readable and computable
The project converges on a simple idea. Useful knowledge should be understandable to a human and interpretable by a machine without losing its documentary identity.
For a reader, that means being able to open an Entry, read a description, examine references, follow relationships, and situate events in time. For an automated system, it means being able to access structured fields, stable identifiers, explicit relationships, and provenance elements.
The graph therefore becomes more than a directory of AI. It forms a selective map where each object retains its type, each relationship must carry documentary meaning, each event must deserve its place, and each item of information can preserve a supporting anchor.
In a rapidly changing technological landscape, this structure aims less to freeze artificial intelligence than to make its transformations traceable. The function of the graph is not to declare that every object in the field has equal status. It is to preserve their differences, organize their relationships, and make knowledge stable enough to be read, connected, and reused across the LXKeys ecosystem.