We present an ontological model that formalizes expert's knowledge used to build three-dimensional structural geological models. This formalization is driven by our intention of proposing a knowledgebased system for automatic model construction. The proposed ontological model includes aspects about geological features, their representations, and modeling processes. This contribution is part of an ongoing PhD project that aims to develop a new knowledgedriven paradigm for three-dimensional geological modeling, one that automatically interprets and uses expert knowledge. Traditionally, the process of building 3D models is considered an issue of numerically representing expert understanding using modeling systems [1]. These systems are incapable of retracing the mental processes of experts. Thus, geological knowledge is always held by experts during the process, and systems work principally on data that represents only a portion of existing knowledge. Our approach challenges this notion by emphasizing knowledge formalization and the automation of the interpretation process. In particular, we concentrate on the aspect of knowledge formalization. To represent geological and modeling knowledge, we employ ontological models. In this paper, we delineate the motivation, requirements, applications, and structure of the proposed ontological models, along with presenting initial results.
itself is multidisciplinary (interpretation, 3D representation, structural geology). Application of ontologies in such multidisciplinarity raises questions about the use of foundational ontologies, harmonization, and alignment. Finally, because geological modelling has been always a process heavily engaging expert knowledge to overcome a very limited quantity of information, the challenge is to maximize the transfer of expert knowledge into a computable framework.
The project has selected case studies centered on modeling folded rocks from observations and existing theories about geological situations. These case studies drive the requirements for the knowledge framework, the interpretation process, and algorithms. Re-use of existing ontologies is a cornerstone of the approach. Preliminary results include:
Algorithm: we propose a three-block algorithm to implement the proposed formalism, Figure.1. The first block consists of a knowledge manager that describes geological features, their geological properties and how to implement them. The second block applies the formal interpretation process. The last block oversees exploring interpreted features in a representation space having both physical and temporal extensions.
2. Formal Interpretation process: we consider this process to be iterative and incremental as shown in Figure.1. In this process, an unexplained situation is selected, information about it is retrieved, theories about this situation are made, then realized, after that checked. Finally, initial understanding by new information is upgraded. This entire process is implemented in the algorithm using a Deeming wheel for continuous improvement [11] with four subprocesses (Plan, Do, Check, Act). In the proposed formalism, interpreting geological features is basically a process of identifying instances of existing types of features, without discovering new types. 3. Ontologies: current work is focusing on the development of an ontological model that describes the three aspects of the formalism (geological features, interpretation process, and representation aspects) as shown in Figure.2. For describing geological features, we intend to adopt the Geoscience Ontology (GSO) of Brodaric and Richards [12]. The GSO is a modular domain ontology that describes geological aspects and features in three layers. The GSO is consistent in many aspects with other top-level ontologies (e.g. aspects of DOLCE and BFO). To structure knowledge about Processes, Observations, Knowledge, Information, and Models and their representations, we create the POKIMOn ontology. This ontology draws from existing ontologies such as Information Artefact Ontology [13], the semiotic triangle and adapts a few concepts from the Observation and Measurements standard [14]. Finally, to structure knowledge aspects about cognitive processes we create the GeolReasOn ontology. This ontology is designed mainly to reason with existing information during the modeling process. It describes situations that could be encountered and possible actions that could be taken to deal with them.
We summarize the goals, approaches, and preliminary results for a PhD project aimed at developing an ontology-driven approach to 3D geological modeling. The overall aim is to better leverage existing knowledge for improved models, with the work currently in progress.
This work was carried out in the framework on the ANR-JCJC project MaLISSiA (Machine Learning and Interpretation of Sub-Surface Architectures) directed by Gautier Laurent and funded by the French national research agency (ANR) under the number ANR-22-CE56-000101. This work was also supported by the BRGM in the framework of I. Laouici’s PhD and TelluS Program of CNRS/INSU in the framework of the project named Cornerstone.