Terminology graph
New field, and everyone is using words you do not have
Every field guards its entrance with vocabulary. Build a terminology graph from your own indexed papers and you get the terms those authors actually use, connected by how often they appear in the same passage, with the source text behind every link.
Free to start, no card. The terminology graph needs a paid plan.
The reason you are here
You have read the paper twice and still cannot tell which terms are load-bearing
Some words are formal constructs with a literature behind them. Others are casual synonyms the author reached for once. From the outside they look identical, and picking the wrong one in a paragraph tells every reader in the field that you are new here.
A glossary does not help, because it lists definitions without weight or relation. What you need to know is which concepts sit next to each other, which ones carry the argument, and which pairing is a live debate rather than a settled equivalence.
That structure is already in the papers you were told to read. It just is not visible by reading them in sequence.
How to learn a vocabulary quickly
Let the corpus teach you its own terms
Three steps, and the graph itself is free to build.
Index the papers you were told to read
The reading list from your supervisor, the three review articles, the handful of papers everyone in the field cites. Twenty is plenty to start.
Build the terminology graph
It runs in the background for that knowledge base and costs nothing. What comes back is the vocabulary of those specific papers, weighted by how much attention each concept gets.
Read your way outward
Start at a term you half understand, see what it sits next to, and click through to the passages that connect them. This is how a field's shape becomes legible.
In the product
The explorer, and the passages behind every link
Search a term, open its neighbourhood, and read the source text that connects it to the terms around it.
What you can rely on
What the graph is, and is not
Extracted from your corpus, not a dictionary
Terms come from the documents you indexed, connected by how often they are discussed in the same passage. It describes your literature, not the field in general.
Weight means something
Node size and edge thickness reflect how substantively your sources treat each concept and each connection, so the graph shows emphasis rather than mere presence.
Follow concept chains
Search any term, view its neighbourhood, and double-click to walk outward through the corpus. Click a term or a connection to read the exact passages behind it.
The agent uses it too
During a workflow the research agent consults the graph to resolve what a polysemous or evolving term means in your specific domain, rather than assuming the general usage.
Building it is free
The graph costs no credits to build and runs in the background per knowledge base.
A paid-plan feature
The terminology graph is not included on the free plan. Indexing, semantic search, the reader and the research agent are, so you can build the corpus first.
What Agent Bayes does not do
- It does not define terms for you. It shows how your sources use and connect them, and the passages are there so you can read the definition in context.
- It is not an ontology and there is no schema to maintain. If you need a formal controlled vocabulary, this is not that.
- It is not on the free plan. You can build the corpus for free and add the graph with a paid tier.
The graph is descriptive, not authoritative. It tells you what these authors do, which is exactly what you need when you are the newcomer.
Before you ask
Questions researchers ask first
How is this different from a keyword cloud?
A cloud counts occurrences. This graph connects terms by how often they are discussed together in the same passage, weights both nodes and edges by how substantively your sources treat them, and links every connection back to the passages behind it.
Is it a knowledge graph?
Not in the enterprise sense, and we avoid the phrase. There is no ontology to define and no schema to maintain. It is a map of how your own documents use their terms.
Does it help the agent or just me?
Both. You use it to orient. The research agent consults it during workflows to resolve what an ambiguous term means in your domain, which is where a general model would otherwise guess wrong.
What if my field's terminology has shifted over decades?
That shows up. Terms that once travelled together and no longer do leave a visible trace, and clicking a connection shows you the passages and their years.
Which plans include it?
It is a paid-plan feature. On the free plan you can still index papers, run semantic search and use the research agent, so you can build the corpus and add the graph when you upgrade.
Related
If you are also wrestling with
Create your account
Index a reading list and see its shape
Start free, index the papers you were told to read, and run semantic search across them. Add the terminology graph when you want the map of how they talk.
- No credit card needed
- The graph costs no credits to build
- Every link opens its source passage