Understand shared meaning

    Why this matters, what we are building and how your practical knowledge can improve it. Start with a question from property work, not ontology code.

    Watercolour of a resident and practitioner clarifying which entrance and dwelling a property report concerns.

    You already know some of the important questions

    Which dwelling does this report describe? What does a quoted floor area include? Does this person participate as borrower in this arrangement, or in another one? These are ordinary questions in property work. They are also questions a useful model must help people answer without relying on a private explanation of what a field name happens to mean.

    This part of the field guide is for people who understand the work but do not practise ontology engineering. You may be a practitioner, data steward, analyst, developer or policy specialist. You do not need formal syntax to identify an ambiguous definition, a missing subject or an unsupported assumption. That judgement is part of how a useful agreement is built.

    Five questions to explore

    1. Why does shared meaning matter?

      See why a correctly formatted value can answer the wrong question. Read the same fictional account as a form, a nested description and connected statements.

    2. What are we building?

      Inspect six real teaching specimens: a glossary entry, dictionary element, taxonomy, controlled vocabulary, resource definition and relationship definition.

    3. What are the benefits and limits?

      Compare an exchange that preserves a useful distinction with one that loses it. Examine the costs and reasonable objections, including schemas, identifiers and AI.

    4. How do the distinctions work together?

      Follow Harbour Court's inspection, report correction and proposed sale. Use supplied evidence to reach a conclusion, then see where a harder case needs more information.

    5. How is the modelling work done?

      See how questions and authorised evidence inform an assisted draft, human judgement and a visible revision. Find what you can contribute before a candidate exists.

    Read in this order if the idea is new to you, or enter at the question you already have. These are reference chapters, not enrolment stages. If you have a candidate page open now, the candidate-reading guide is a direct starting point. If you are making formal representation decisions, choose Ontology modelling.

    A small fictional case, with room for difficult questions

    Harbour Court contains Flat 1 and Flat 2. One inspection of Flat 1 leads to two report versions. A correction clarifies an access limitation without creating another visit. Separate area accounts differ because they include different spaces. These supplied facts let you practise reasoning from evidence, not guess an answer from a familiar date or address.

    The story is not customer data or an implementation of the current Property Pack candidate. Other chapters vary the subject and situation so that you can apply the distinctions beyond Harbour Court. Worked responses stay visible; compare the reasoning with your own, including what each answer leaves open.

    Questions you may have before starting

    Is this another dictionary?

    A dictionary is one useful view. The agreement also describes subjects, relationships, topics and choices. See the six aligned outputs to understand what each lets you inspect.

    Will we have to replace our systems or share one database?

    Agreeing meaning does not inherently require either. Existing forms and systems can remain useful, but implementation and migration still require separate decisions and checks. Read the conditions and limits.

    Why not use one definition or an existing standard?

    Reuse helps when meanings fit. Some local differences are necessary, and a spelling match is not evidence of equivalence. Compare the case for shared and contextual definitions, including how existing standards inform the work.

    Does a model guarantee accurate, safe information?

    No. It can describe evidence and relevant handling distinctions; it cannot make a source true or grant somebody access to records. Accuracy and access need separate evidence and responsibilities.

    Is an AI proposal already an approved standard?

    No. Drafting, technical checks, human review, authorised release and external recognition are distinct. Read the status of the particular artefact and see who judges what. A reasoned unresolved question is more useful than an unsupported agreement.

    Comments

    Loading comments…