Glossary — the vocabulary, in plain language#

Two sentences per term, no prerequisites. For the full story behind any of them, follow the link at the end of the entry.

Linked Data: Data published so that every thing (a cell, a material, a test) has a web address, and records point at each other by those addresses instead of by ambiguous names. It is what lets someone else’s software follow your cell → test → dataset chain without guessing.
IRI: The web address that names one record, e.g. https://w3id.org/battinfo/cell/gm6p-mrc6-p6cc-0m18 — an internationalized URL used as a permanent identifier. BattINFO IRIs never change and never get reused, so they are safe to print on a cell wrapper or cite in a paper (Identifiers).
EMMO: The Elementary Multiperspective Material Ontology — a big, curated dictionary of science and engineering concepts, each with a formal definition and an IRI. BattINFO uses its battery and electrochemistry branches so that “capacity” in your record means exactly the same thing as in everyone else’s.
Ontology: A machine-readable dictionary plus the rules relating its terms (“a coin cell is a battery cell”). Software uses it to reason about your records; you mostly never touch it directly.
JSON-LD: Ordinary JSON with a small @context block that maps its keys to ontology IRIs — human-readable and machine-interpretable at once. BattINFO generates it from your records automatically; you author plain JSON or Python.
RDF: The underlying data model of Linked Data: every fact is a triple (subject – property – value), and a pile of triples forms a graph. JSON-LD is one way of writing RDF down; Turtle is another.
SPARQL: The query language for RDF graphs — SQL’s cousin for linked records. It is what makes questions like “all datasets from NMC811 cells cycled at 45 °C” answerable across datasets that have never seen each other.
spec vs. instance: A spec is a reusable description — the datasheet: a material grade, a cell product, a test protocol. An instance is one physical realization — this jar of powder, the cell labelled B7-01, the test run that finished Tuesday.
descriptor: The research-grade level of a cell spec: electrode composition, electrolyte formulation, separator, construction — the detail below the datasheet numbers (Guide 5).
@type stacking: One record carrying several ontology classes at once: a cylindrical LFP cell is simultaneously a BatteryCell, a CylindricalBattery, and a LithiumIronPhosphateBattery. BattINFO stacks these automatically from the format and chemistry you enter.
SHACL: A rule language for checking RDF graphs — the semantic layer’s counterpart of JSON Schema. One of the layers behind validate_record_report (validation contract).
content negotiation: One IRI, several representations: the same address returns JSON-LD to a script asking for application/ld+json and a human web page to a browser. This is why an IRI can “open a sign-in page” for you but still serve data to machines (troubleshooting).
BDF: Battery Data Format — the one tidy, documented table layout that every cycler export is converted into (ws.convert() / ws.convert_csv()), so downstream tools read one format instead of ten.
short ID: The first six characters of a record’s UID (e.g. gm6pmr) — the handwriting-sized version of an IRI, used in filenames and on cell labels (Label your cells).