aiExpert In Practice – #1 – Aeronautical Data model

Source Model

The source of this model was the AIXM 5.2 Aeronautical Information Exchange Model, owned by Eurocontrol and the FAA. The model was downloaded from https://aixm.aero/document/aixm-52-data-model-uml and is used without permission, as we believe this model is public domain.

The purpose of this document is to describe the results returned when parts of this model were analyzed by the aiExpert ‘Static Data’ query, which is used to examine all kinds of static class models, logical-  and conceptual- data models.

Please note that this is using the out-of-the-box function of aiExpert, with no tailoring of the prompts to suit the domain.

Model Diagram: 'Airspace'

We took the diagram ‘Airspace’ (just because this seems fairly important given the domain) :

As you can see, this is a complex bit of modelling

Inputs

We chose the ‘Static Model check’ query for this analysis, which collects all the data about each class, all its attributes and connectors, and submits this ot the LLM with the prompts:

System instructions: “You are an expert in the structure of static data or class models”

Rules: “Here are some class names and their connections which are part of a static model.
Check if they make sense according to their names, roles and connections..
When you respond, grade the severity from 1 – a minor issue, to 5 – a serious problem.
Always include the ID of each element with any feedback ”

LLM used: gpt-5-mini

Data collector: ConnectorDetail

Important note:

We present the details of the Ai analysis largely without comment. This is not our model: we don’t understand the domain, so we are not in a position to say whether the feedback from the LLM is good or bad, useful or useless.  What we can say is whether the feedback might be useful, or whether it seems silly.

In the ‘suggested fix’ information which is returned by the AI, we highlight in italics the part of the input message to which the feedback refers.

Observations

As with all feedback from aiExpert, this is the opinion of the LLM at a point in time. We are not in a position to say what’s right, but all the feedback (there were two more, omitted to keep this article short) seems to be making a point which refers to some aspect of the model. If the authors of the model, or anyone else with knowledge of the domain knows better, please let us know.