From Observation to Data: A Framework for Capturing Reality

Disclaimer: According to Immanuel Kant, we can never observe reality as “Ding an sich” or noumenon. Our process always begins with Observation: the act of perceiving phenomena—the objects, events, or signals that appear to our senses or instruments, before we have assigned them a formal meaning.

From Observation to Recording

While we can observe countless phenomena, we cannot map, enumerate, or analyse them in a structured way without an Ontology defined as as Domain of Discource or at least as a Univers of Discource (See phase 1 and 2 of the Design rational). An ontology is the formal “rulebook” that allows us to move from “unstructured” observation to “structured” Recording—the process of creating data.

The concepts of “unstructured” observation and “structured” Recording are at each end of a continuum defined by how well-defined (scoped) our ontology is. In practice, we rarely operate at the extreme ends of this continuum, but rather somewhere in between, depending on the project’s purpose and current state.

Large projects with many participants, running over more extended periods, typically require more well-defined ontologies to ensure consistency between individuals and over time, compared to a single-person, one-off survey. Strictly speaking, we are not even able to verbalise our observations without a loose, culturally defined ontology.

Even a well-thought-through ontology (DoD), rarely survives contact with the complexity of observed phenomena. This is why a pilot study 🧪 is essential. It is an iterative process where you test your recording rules in the field, discovering ambiguities and refining your schema. This crucial step ensures that your final ontology is not only theoretically sound but also practically robust, enabling a consistent and meaningful recording of your observations. Far too many projects suffer from having to change their ontology after data recording has started, and find themselves in the dilemma of having to scrap and redo the first part of the data recording or try (seldom successfully to modify the observations to match the new ontology.)

The SemanticGIS approach to generating New data

1. Precompilng a ontology: Desktop work where a preliminary ontology is created based on a priori knowledge and literature studies.
2. Pilot registration: The purpose is to test the the registration based on the preliminary ontology on “reality” in the Area Of Interest. We often use “unstructured” registration tools using paper and notebooks to ensure maximum flexibility in the registration.
3. Finalising the Ontology: Modifying the preliminary ontology to match “reality”
4. Creating a data schema: Specifying how data is to be stored in a computer system
5. Final Registration: A new registration using a more structures tools i.e. apps preprinted forms etc.

In SemanticGIS Design the Preliminary Ontology is as part of the Project scoping and Data Modelling phases og the project and documented in ( step 1 and 2) of Design rational. The following steps all belong to the Data Sourcing and Preparation phase of the of the project.. In both the Pilot and the Final Registration, the same four fundamental methods of data capture are used. The crucial difference lies in the goal of the work, the tools you employ, and the level of rigor you apply. The pilot is an exploratory test of your ontology using flexible tools, while the final registration is a systematic execution of your finalized rules using structured tools to produce clean, consistent data.