The knowledge graph
Seeing how records are connected, what is about the same thing by meaning, what stops if something fails, and what could cause a fault.
The detailed manual is currently available in English.
The Knowledge graph (side bar) shows records and their relations: what powers what, what is part of what, which IOC acts on which equipment, which documents apply. Beside it, the semantic graph shows what is about the same thing, from what is written, even where nobody linked it.
Explore
- Open Knowledge graph and pick a record (or open it from a record’s page).
- Centre here on any record to move around. Inbound and Outbound show the two directions of a relation.
- Choose the layers (control, power, cooling, vacuum, timing…) to follow only those kinds of relation.
Related by meaning (the semantic graph)
ARGUS indexes what is written: procedures, tickets and their comments, comments on equipment, and the text of attached files (see AI settings). For a piece of equipment, a ticket or a document, Related by meaning finds the other records whose written passages are closest in meaning to its own, in any language:
- each with how close it is, from 0 to 1, and why: the two passages, its and yours, that are closest;
- only what you may read: restricted records, other workspaces’ private documents and riservato documents never appear;
- equipment with nothing written about it yet is compared by its key, name, type and description.
Where to see it:
- on the web, the Related by meaning card on an asset’s, a ticket’s and a document’s page, and the related by meaning switch in an asset’s relation graph, which adds them as a column joined by dashed lines;
- in the mobile app, the Graph tab (start from equipment, a ticket or a document: what it is connected to, and what is related by meaning; tap a neighbour to walk on), and the Related by meaning section of each record.
A link by meaning was found, not made by anyone: it is a lead to check, such as an earlier fault fixed the same way or a procedure that applies, not a fact. What has not been indexed yet cannot be found: the index is updated when a document is published and on a schedule (AI settings).
What stops if it fails?
Pick a record and ask What stops if it fails?: ARGUS follows the recorded dependencies the way a failure travels and lists what could be affected and by which path. Losing a power supply and losing an IOC have different effects: a magnet stops without its supply, while devices an IOC reads keep working, unread.
It misbehaves: find causes
For a fault, use It misbehaves: find causes: add what misbehaves and, if you know, what still works. ARGUS walks back to the records that could explain all of it, most likely first.
Ask ARGUS can do both in words: “what depends on the chiller CHL01?”, “the gun vacuum and the RF both tripped: what could cause it?”.
Read the results with care
The graph knows only the relations that were recorded: a missing or wrong relation gives an incomplete answer. A suggested cause is a candidate to investigate, not a diagnosis, and its ranking is an order, not a measured probability. A record that does not appear is not thereby unaffected.
Open the records behind each result and compare them with what you observe, the tickets and the procedures. The graph describes how things are connected, not how the machine is running right now.