A semantic data model for mineral and energy resource exploration
Definitions
Ontology: An ontology is the definition and
classification of concepts and entities and the relationships
between them. Semantic data: Semantic data is data that is
organised so it can be understood by machines. Exploration:The process by which geological
information is collected and analysed to identify mineral or energy
resources as well as determining the economic feasibility of their
extraction.
An ontological approach can help both humans and computers to understand, integrate, and analyse
exploration data across
a challenging variety of data. Change factors include:
The accelerating volume and variety of data types, data formats, and data sources.
The use of big data, machine learning and AI to find new discoveries. Computers need semantic
data.
The incorporation of data from other domains, such as life sciences data, into exploration data
holdings.
The expectation of real-time data streaming, processing, and analysis.
The move away from relational databases to NoSQL databases, document databases, and object
storage.
Using the ontology in exploration
Explorers undertake a range of activities to
understand the geological properties of a
geological or administrative feature. The process
typically goes like this:
We undertake a survey on the feature at a
site.
The site may comprise of the whole feature, part of the
feature,
or may encompass and extend beyond the feature.
The survey yields samples that may be physical, such as a
drillcore, or non-physical proxies such as photographs.
We conduct observations on the samples using various
procedures.
The observation yields results as measured values or
qualitative
descriptions.
The results are interpreted to understand the geological
properties of the feature, e.g. mineralogy or presence of
hydrocarbons.
Examples of applying the ontology
The benefit of the ontology is that it can be applied across a wide variety of exploration methods.
For example:
An entity or location within, or encompassing, a feature that acts as a proxy to represent a
complete (ultimate) feature.
A Feature of Interest is proximate when it represents a larger feature, as opposed to being a
discrete component of a larger feature. e.g. an outcrop can be examined as a representative of a
formation, whereas a formation does not represent a whole basin but is a component of it.
Where a sampling is undertaken, but the sampling geometry and site geometry do not necessarily
have to be equivalent.
The sample is a representative part of a feature of interest.
Samples may be original samples, subsamples where a new sample
is split into smaller samples, processed samples where a sample content is
retained but is processed to have altered properties, or duplicates - identical
samples.
A sample may be surveyed to produce a new sample or sub-sample.
Examples: drill core, drill cuttings, soil sample, hand specimens, water, photograph, LAS
file.
Observation
An act of carrying out an observation using a procedure to measure, estimate, calculate
a value of, or describe a feature, site or sample.
Observations differ from sampling in that sampling yields an artefact, whereas an observation
yields a qualitative or quantitative result.
Observations may be the observation of the physical limits of an interval.
The result of the observation performed on a sample, stored as a description or as a value and
unit of measure.
Examples:
Physical properties, e.g. concentration, mass, temperature
Petrographic descriptions
Geophysical measurements e.g. gravity, magnetic field strength
Petrophysical log measurements e.g. gamma, density, resistivity.
What are the business advantages of an ontology?
Ontologies, along with taxonomies and vocabularies, provide these business advantages:
The ontology provides the business with a shared understanding of business concepts, entities,
and relationships.
People across business divisions, as well as customers and suppliers, can use the ontology as the basis for
seamless information exchange.
Ontologies and their vocabularies allow multi-language taxonomies, and synonyms, regional, historical and colloquial
terms for concepts. This enables semantic master data management.
Teaching the ontology to new recruits during induction will help them to understand the
business.
How does a computer understand the ontology?
The ontology is made available to the computer in the Web Ontology Language (OWL) as an RDF (Resource
Description Framework) file.
The computer can understand this semantic description of the data entities, their attributes, and
their relationships.
We use RDF-based controlled vocabularies to feed the computer the taxonomy - a list of words related
to each other. The computer uses SPARQL query language to query the vocabulary API to understand the
words and their meaning.
A borehole example of computer reasoning
This example demonstrates the use of semantic data techniques that enable the computer to
reason (understand) without needing a human to tell it what to do.
We feed the computer data for borehole CARINYA SOUTH 3 using the Persistent
Identifier (PID) BH063772.
The computer can reason that a borehole is a type of geological
site by querying the Geological Properties Ontology.
The computer can reason that a Borehole has alternative labels of Core Hole,
Corehole, Drillhole, and Well by querying the borehole concept in the geological sites vocabulary.
The computer can reason from the borehole purpose that it will process this borehole as a
Coal Seam Gas Well.
Sure, we could have told the computer straight up that this was a Coal Seam Gas Well.
However, hopefully you can see the power of enabling the computer to reason across Features, Sites,
Surveys, Samples, Observations, and Results, and all of their attributes, relationships, ontologies
and vocabularies.
What about the existing data models for exploration and appraisal data?
Standards such as
GeoSciML
(Geoscience Markup Language) for minerals and
PPDM
for petroleum and gas are detailed data models that fit within the
ontology model and inform it.
Many of the geoscience data models exist as relational data models,
and lack the semantic information required by computers for activities such as machine learning.
This ontology is not about replacing, but instead complements, extends and integrates these data
models.
The ontology enables integration of data across these different models. For example, if we integrate GeoSciML
mineral data with PPDM petroleum and gas data, the computer will know that a borehole and a well are
synonymous.
Where does my database fit in?
This ontology can be represented in a relational database. However, if you're starting afresh, you
should consider databases that support key-value data or document (JSON-based) data structures.
Graph databases are perfect for OWL and RDF data.
Here are some things you can do to make your existing relational database more semantic:
Map your existing reference data tables to ontologies and vocabularies.
Create and store persistent identifiers for both data and metadata.
Create semantic data views so machine consumers can query these views.
More information
We hope you enjoyed this introduction to the Geological Properties Ontology. For further information please contact us at info@crosslateral.com.au.
This is an open-access article distributed under the terms of the
Creative Commons Attribution 4.0 International License, which permits
unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited. For a full description
of the license, please visit
https://creativecommons.org/licenses/by/4.0/legalcode.