Ambiguity and ORM2 (Object-Role Modeling v2)

ORM2 is a type of fact-based modelling

Victor Morgante
12 min readMar 17


2023–03–18 — Resubmitted in full, with further notes for genuine researchers. Original of this article at 2023–03–01 on GitHub: here

ORM2 (Object-Role Modeling v2) is a type of fact-based modeling, which is a type of concept Role Modeling is a graphical language.

The short answer to the question posed in the title of this article is:

“Yes, under certain circumstances, sentences of ORM2 can be constructed, the interpretation of which are ambiguous”.

This might be contrary to marketing material you have read on Object-Role Modeling, or indeed contrary to academic papers published on Object-Role Modeling and by journals of varying reputation.

Things are not entirely lost for the application or utility of the ORM notation, only to any claim to the efficacy of ORM under the formalisation of ORM. It would be dishonest for my company to produce ORM based software without having first disclosed the limitations of ORM outside of software.

Consider the following ORM2 diagram:

The symbol in the middle of the diagram is a Join Subset Constraint. It looks like this:

A Join Subset Constraint joins a set of roles played by ostensible objects in a subset / superset relationship.

From the model above, it is ambiguous as to whether the Join Subset Constraint says which of the following:

or this….

So basically, an interpreter of that diagram would not be able to distinguish as to whether a booking was made for a seat in a row in a cinema where that was not allowed by the cinema, or whether the booking was made for a seat ina row at a cinema that merely contains that row.

NB See the bottom of this article where we cover a case where we say that Fact Type Readings cannot have negation.

Clearly this is not an ideal situation for a visual language promoted as formalised, as in logically formalised?

Update 2020/09/13: More succinctly, a set of well formed sentences of ORM may be constructed, the interpretation of which is ambiguous. While true of any…



Victor Morgante

@FactEngine_AI. Manager, Architect, Data Scientist, Researcher at


See more recommendations