Overview
The vocabulary and the core ideas behind Clarity: what it knows and how it keeps knowing it.
Clarity is Motif's financial research engine. Underneath it is one durable structure, a graph of what the market is doing and why, plus a handful of ideas that keep that graph accurate over time.
This section defines those ideas. It is the conceptual reference: what each thing is and the rules that govern it. For the step-by-step flow of how a document becomes an answer, see the Pipeline.
The ontology
Clarity's ontology is the shape of what it knows. Everything it ingests is reduced to a small set of object types and the links between them, so a news article, a filing, and a transcript all land in the same structure.
Entity
The things news is about: companies, people, places, events, and topics. Each carries aliases (abbreviations, tickers, acronyms) that let later documents match the same entity. Entities fall into a few broad categories: actors, assets, events, places, and concepts.
Claim
One atomic fact about an entity, such as a quote, a figure, or an event. It carries the source it came from and the window of time it was true. Facts are kept atomic so each can be cited, dated, and superseded on its own.
Relationship
A directional, typed link between two entities. Most are context (structural, like "supplies" or "is CEO of"). Some are causal ("X drove Y"), carrying a strength, lag, and direction so the chain can be followed both ways.
Story
A time-bounded narrative arc that groups related entities and developments, such as a leadership transition over a quarter or a supply shock and its fallout. One entity can sit in several stories at once.
Theme
A broader roll-up of related stories, the connective tissue that lets an answer bridge from one arc to a neighbouring one.
Asset
A tradable instrument, an equity or a supported crypto token, that an entity is linked to. This is how a story about a company resolves to a ticker you can hold.
Source
The original document a fact was extracted from: a news article, filing, transcript, or manual upload, carrying its title, publisher, URL, and publication date. Every claim and relationship is sourced from exactly one original article, and entities and asset links carry the same provenance. This is the evidence that grounds every fact, and that link is what each inline citation resolves to, so any fact in an answer can be opened back to the article it came from. Nothing enters the graph without a source.
Two properties cut across all of these:
Time. Claims, relationships, and stories carry validity windows, so Clarity can answer both "what is true now" and "what was true then." A changed fact is superseded: its predecessor is stamped with an end date rather than overwritten, so history stays intact.
Salience. Clarity tracks how central each entity is to the graph, so answers lead with what matters rather than whatever matched the words.
Live prices are not part of the ontology. They are overlaid at question time, so commentary stays current without the graph going stale.
Causality
Relationships are the part of the ontology that makes Clarity more than a search index, and they come in two kinds.
Context is a structural tie: "supplies", "is CEO of", "is headquartered in". It says how two things are related.
Causal is an assertion that one thing moves another: "the rate cut drove the rally", "the strait closure lifted freight rates". A causal link carries a strength, a lag, and a direction, and it is always oriented from the driver to the thing it affects, no matter how the sentence was phrased.
Because the direction lives on the link rather than in a name, a causal chain can be traversed both ways: forward to answer what does this affect next? and backward to answer why did this happen? That is what lets Clarity follow a shock from geopolitics through a commodity to opposite-signed moves in two stocks and surface it as a diversification signal, instead of just listing facts that happen to mention the same company.
When several drivers converge on one outcome, Clarity keeps the chain connected end to end so no step drops out of a later question. Extraction covers how a causal link is oriented and how a multi-step chain is completed.
The lifecycle
Four ideas keep that ontology populated and trustworthy. Each has its own page.
Ingestion: how source material gets in. The feeds, the staging, and the screening that keeps noise out, all off the request path.
Extraction: how one document becomes structured ontology objects. Entities, the facts about them, and the relationships between them, including cause and effect.
Memory: what Clarity keeps and answers from. A curated, durable graph, kept separate from raw reading and from the answers it produces.
Asset linking: how an entity is connected to the equity or token you can trade, so news reaches holdings.
Dreaming: the background self-curation that keeps memory concise and connected over time. It merges duplicates, retires stale facts, and re-ranks what matters.
Last updated