Plain-English glossary

Every word in the guide, in plain English

If a term anywhere in the Learn section made you pause, it's defined here — each one a friendly sentence or two, plus a tiny example or analogy to make it stick.

Terms are grouped by topic below. To jump straight to one, start typing in the box and the list filters as you go.

The building blocks

• Graph — a web of things and the connections between them; the "whiteboard of dots and strings." Unlike a spreadsheet of rows, a graph treats the connections as first-class data you can follow and search.
• Node — one "thing" in the graph (a person, account, product). If the graph is a whiteboard, a node is one of the dots you draw.
• Edge / relationship — a labeled, directed link between two nodes ("Ana KNOWS Ben"). It's the string tied between two dots, with an arrow showing which way it points.
• Property — a piece of data attached to a node or edge (a name, a date, an amount). Example: the "Ana" node might carry {age: 31}, and a SENT edge might carry {amount: 4200}.
• Label — the kind of thing a node is (Person, Account), so you can ask for "all the people" at once. Like a colored sticky note that sorts dots into types.
• Path — a chain of nodes and edges from one point to another; the "route" through the graph. "Ana → Ben → Cleo" is a path of length two.
• Traversal — the act of following edges from node to node to answer a question. "Friends of friends" is a two-hop traversal — you step out, then step out again.
• Hop — a single step across one edge. "Two hops out" means you followed two relationships in a row.
• Knowledge graph — a graph that captures facts about the real world — people, places, products, and how they relate — so software (and AI) can reason over them. Think of it as a fact-checked map of a domain.
• Schema — the optional "shape rules" for your data: which labels exist, what properties they carry, how they connect. A graph can be flexible about this, which makes it easy to add new kinds of things later.
• Index — a behind-the-scenes lookup structure that lets the engine jump straight to the data it needs instead of scanning everything. Like the index at the back of a book — it turns "search the whole thing" into "flip to page 214."

Meaning, vectors, and retrieval

• Vector — a long list of numbers that acts as a "fingerprint" of meaning, so similar things land near each other in number-space. Two recipes for the same dish get similar fingerprints even if they share no words.
• Embedding — the specific vector produced for one item (a document, an image, a record) to capture what it's about. "Make an embedding of this paragraph" means "turn its meaning into numbers."
• Vector search — finding the items whose vectors are closest to a target vector. It's how a computer answers "what's most similar to this?" at scale.
• Semantic search — finding things by meaning rather than exact keywords ("show me records like this one"). A keyword search wants the same words; a semantic search wants the same idea.
• RAG — "retrieval-augmented generation": before an AI answers, you fetch the most relevant facts and hand them over, so it responds from real data instead of guessing. Like giving someone the open book before the exam.
• Retrieval gap — the slowdown and fragility you get when a single real-time answer has to hop across several separate databases. The question is simple; the plumbing is what makes it slow.
• Polyglot persistence — the common practice of using several different databases together, each for the job it's best at. It's flexible, but it's also exactly the "database zoo" that creates the retrieval gap.

Languages and patterns

• Query — a question you ask the database, written in its query language. "Give me Ana's friends-of-friends who like jazz" is a query.
• Cypher — a popular, readable language for asking graph questions; the query tends to look like the question. You draw the shape you want with little arrows, and the engine finds it.
• GFQL — a second, data-frame-style query language xrayGraphDB also speaks, handy for pipeline-style work where you describe a chain of hops as a sequence of steps. People from notebook and data-frame backgrounds often find it natural.
• MATCH — the part of a Cypher query that says "go find this shape." It's where you sketch the pattern of dots and strings you're looking for.
• RETURN — the part of a query that says "and give me back exactly this" — a name, a count, or a whole path.
• Pattern — the shape you're searching for, written as nodes and edges. A graph query describes a destination (the pattern), not turn-by-turn directions for finding it.
• Self-documenting — the database carries its own manual inside it; run SHOW FUNCTIONS and it lists every built-in capability with its name, category, and a plain-English description — always current, because it comes straight from the engine itself, not a separate manual that could drift out of date.
• SHOW FUNCTIONS / SHOW PROCEDURES — the live commands that ask xrayGraphDB to describe itself. SHOW FUNCTIONS returns every built-in function with its name, category, and description. SHOW PROCEDURES does the same for procedures. The results are the authoritative catalog; what the engine says it can do is exactly what it can do.

Finding structure and patterns

• Graph analytics — algorithms that reveal the hidden structure of a network — who's important, who clusters together, what's the shortest route. They run inside xrayGraphDB, right next to the data.
• PageRank — an algorithm that ranks the most important or influential nodes in a network, by counting not just how many connections a node has but how important those connections are. Famously how early web search ranked pages.
• Centrality — a family of measures of how central or well-connected a node is — the hubs, brokers, and key connectors. The "who's a big deal here?" score.
• Community detection — finding clusters of closely-connected things (customer segments, friend groups, suspicious rings). It's how you spot the natural neighborhoods in a network nobody labeled in advance.
• Shortest path — the most direct route between two nodes. The classic "six degrees of separation" question: how few hops link two people?
• Connectivity — whether (and how strongly) parts of the graph are reachable from one another — which things form one connected whole and which are islands.

What makes it fast

• Vectorized — an engine design that processes thousands of records in one sweep instead of one at a time. Like a checkout that scans a whole basket in one pass rather than ringing up items one…by…one. This is xrayGraphDB's primary engine.
• SIMD — "single instruction, multiple data": doing the same operation on many values at once, a key hardware trick behind vectorized speed. One instruction, lots of numbers crunched together.
• GPU / GPU-accelerated — using your graphics card — a chip built to do thousands of small calculations in parallel — to finish heavy graph algorithms in a flash. xrayGraphDB falls back gracefully to the CPU when no GPU is present, so nothing breaks if you don't have one.
• Warm query — a query whose data is already loaded and ready in memory, so it returns at top speed (about 0.2 ms in xrayGraphDB). A "cold" query has to fetch its data first and is naturally slower.
• QPS — "queries per second": how many questions the database can answer each second. Higher is better; xrayGraphDB sustains thousands when pipelined.
• Pipelined — sending the next request without waiting for the previous answer to come all the way back, so the engine stays busy. Like a relay where the next runner is already moving — it's how throughput climbs to thousands of QPS.
• Latency — how long one answer takes to come back, usually measured in milliseconds. It's the "how fast did this question feel?" number, as opposed to QPS, which is "how many per second overall?"

Trust, isolation, and tenants

• Transaction — a group of changes that all succeed or all fail together, never half-done. Like a bank transfer: the money leaves one account and arrives in the other, or neither happens.
• Multi-tenant — serving many customers (tenants) from one system, with each one's data kept cleanly separated. One building, many locked apartments.
• Isolation — the guarantee that one tenant can never see or touch another tenant's data. In xrayGraphDB this is always-on, enforced by the engine, not by hopeful application code.
• Key (encryption key) — the secret that locks and unlocks encrypted data. Without the right key, the stored bytes are unreadable noise. Each tenant gets its own independent key.

Protection terms

• Encryption — scrambling data with a key so that anyone without the key sees only meaningless noise. The math is easy to do with the key and effectively impossible to undo without it.
• Encryption at rest — your stored data is kept scrambled on disk, so a stolen drive or backup is useless without the key. xrayGraphDB does this natively, per tenant, always on, with no off switch.
• AES-256 — a widely trusted, strong encryption algorithm. The "256" is the key size — large enough that brute-forcing it is far beyond any practical computer. It's the modern default for serious data protection.
• AES-256-GCM — the specific flavor of AES-256 xrayGraphDB uses. The "GCM" part also detects tampering, so you know the data wasn't quietly altered. It's the same algorithm the U.S. NSA approves for information classified up to Top Secret.
• Key (encryption key) — the secret that locks and unlocks encrypted data. Guard the key and the data is safe even if the storage is stolen.
• Per-tenant key isolation — each tenant's data is protected with its own independent key, kept separate from every other tenant's. Even if one key were exposed, it couldn't unlock anyone else's data.
• FIPS — the U.S. government's standard for validated cryptography — meaning the crypto has been independently tested to behave correctly, not just claimed to. xrayGraphDB's crypto meets FIPS 140-2.
• CNSA — the modern suite of cryptography the U.S. government specifies for protecting sensitive and classified national-security information. xrayGraphDB's architecture is designed to meet that high-assurance bar, with formal accreditation in progress.
• NSA — the U.S. National Security Agency, the body that sets the cryptographic bar for protecting government secrets. When we say AES-256 is "NSA-approved for Top Secret," that's the approval being referenced.
• "Top Secret" — the U.S. government's classification for information whose exposure would cause exceptionally grave damage. We use it only to name the algorithm tier AES-256 is approved for — not to claim the product itself is certified for handling Top-Secret data.
• SOC 2 / HIPAA — widely-recognized compliance frameworks: SOC 2 for how a service handles security and customer data, HIPAA for protecting health information. xrayGraphDB meets their requirements.

Note: the architecture is designed for classified, high-assurance environments with formal accreditation in progress — we don't claim the product is certified or accredited for Top Secret.