Understanding the High Value of Ontologies for Natural Resource Exploration

Companies recognize the value of the data captured through their history.  However, data is being collected in the last decades and organized under several different models conceived to attend tasks and applications. Besides structured data, documents, spreadsheets and images complement the inventory of strategic corporate information.

The situation of most large petroleum companies today is hundreds, maybe thousands of database tables supporting hundreds of applications and more than thousands of documents spread in virtual drawers. A set of tables reflects the process of well drilling, while another set describes reservoir characterization. More tables organize the information about reservoir quality, which is added to several proprietary files of reservoir modeling and analysis. Technical reports summarize the stratigraphic interpretation of the sedimentary basin and production reports register the amount of oil produced along with the expenses in facilities and maintenance.

Then a manager wishes to answer common questions like “Why the well 1 is producing less than the well 2?”, “Why the production decrease in well 3?”, “What are the factors that made the play X demands more maintenance than usual?”. The answers are there, into the data, submerged and hidden under models and applications that reflect different views over the information.

Then comes the big question: How can I extract some useful information from this data swamp?

Ontology is our response. Ontology is a theory based on Philosophy that helps in making explicit the intended meaning of a vocabulary according to a conceptualization of a community of stakeholders. When applied to build models that express this vocabulary, an ontology builds an artifact (an RDF or OWL file, or a structured document) that specifies formally the set of shared concepts used in the community. Ontology artifacts are machine-readable, which means that the explicit represented semantic can be used by computers to solve problems with no human intervention.

Ontologies are useful in at least three ways: 

  1. Standardize the way in which data is collecting, avoiding free reports that are hardly processed by computers;
  2. Brings out the semantics of stored legacy documents, allowing them to be integrated to support data analytics and insight engines.
  3. Indexing visual content of images, photographs and figures in old pdf data.

A good example of the situation (1) can be seen when comparing Figures 2 and 3. Figure 2 shows a description of a columnar section of Aruma Formation of Saudi Arabia [Al-Kahtany, El-Sorogy et al. 2016] and Figure 3 the description of a well in Campus Basin, Brazil. Both sections represent, in a pictorial way, the lithology, textures, sedimentary structures, fossil content and contacts of formations. However, the description in Figure 3 was produced with an ontology-based software, the Strataledge, which controls the sedimentary characterization in the description using a formal vocabulary.  Figure 3 itself was not drawn, but consist of one of the exportation formats that integrate the columnar information along with petrographic data of the rock. Contrasting with the human-made interpretation shown in the last column of Figure 2, the ontology allows that the depositional environment being automatic interpreted by the ontology-based system [Carbonera, Abel et al. 2015].

Figure 2 – Lithostratigraphic section of the Aruma Formation, Khashm Bowaibiyat area, Northeast Riyadh [Al-Kahtany, El-Sorogy et al. 2016].

Figure 3 – Stratigraphic and petrographic integrated study in Campos Basin. 

Although domain ontology plays an important role in standardizing and structuring corporate information, its value is strongly recognized when using to extract correlations of the stored information.  That was the case of applying data mining for the extraction of reservoir petrofacies from petrographic data, captured using Petroledge ontology. The tables exported by the system share the same vocabulary and format for the names of minerals and pores, textural aspects, and all the transformations suffered by the rock through diagenesis. The quantification of all the aspects of the rocks allows applying clustering techniques and feature selection to identify the petrofacies that are related to the level of porosity of the petroleum reservoir rock. The ontology and clustering techniques applied in the ongoing work of [Fernandes 2019] has achieved results close to those of experts doing the same task: 85% of similarity in the separated sets of machine and human expert.

However, the reality on the corporate environment is that the stored data does not share a consensual model based on ontologies and it still needs to be consulted for decision support.  In that case, ontologies can be used to build a semantic layer that represents the meaning of the main concepts of the organization. The role of this layer is to connect the human needs with metadata of structured data and to the content of documents.  Domain ontologies are associated with techniques of knowledge graphs, information retrieval and insight engines to provide a uniform way of automatically indexing legacy data. In the work of [Lopes, Alvarenga et al. 2019 ], the ontologies of Strataledge and Petroledge were applied to index pdf and doc documents. The documents were scientific journals on sedimentology, stratigraphy and technical reports of integrated projects on Sedimentology, Petrology, Seismic Stratigraphy and Biostratigraphy of Campos Basin in Brazil.  The ontology supported the word sense disambiguation for automatic indexing of the terminology of the document for information retrieval and data analytics.

Domain ontology provides very useful support for labeling and indexing visual content. Petroledge ontology supports the Rock Viewer application that captures and indexes photomicrographs of sedimentary rocks, based on detailed petrographic aspects of the rock: mineralogy, texture, crystal habits, contacts, fabric, diagenetic paragenesis and so on.  The software produces an album of commented pictures that can be retrieved and cross-correlated regarding any feature of the rocks that can be associated with reservoir quality.

Regardless of the effort in developing heavy well-founded ontologies, their use has proven to be valued to approximate the intention of the consultation in the mind of users with the meaning of vocabulary dispersed data lakes in corporate environments.   Strataledge ontology is built over 700 structured terms that cover all terminology on facies description for magmatic, metamorphic, chemical, clastic, and metasomatic types of rocks. A complete terminology of lithology, textures, fabric, and structure are specialized for each type of rock class and associated individually with a visual icon that represents each particular aspect.

Petroledge ontology includes more than structured 800 terms for all mineralogical classes, fabric, textures, diagenetic transformation, and structures found in the composition of magmatic, metamorphic, chemical, clastic, and metasomatic types of rocks, including Brazilian pre-salt lithological types.

Strataledge and Petroledge ontologies licenses are available to develop applications and databases.

Visit www.endeeper.com for more information about Endeeper ontologies.


Abel, M., E. S. L. Gastal, et al. (2019). “A knowledge organization system for image classification and retrieval in petroleum exploration domain.”

Al-Kahtany, K. M., A. S. El-Sorogy, et al. (2016). “Stratigraphy and depositional environments of the Upper Cretaceous Aruma Formation, Central Saudi Arabia.” Arabian Journal of Geosciences 9(5): 330.

Carbonera, J. L., M. Abel, et al. (2015). “Visual interpretation of events in petroleum exploration: an approach supported by well-founded ontologies.” Expert Systems With Applications 42(5): 2749-2763.

Fernandes, L. P. (2019). A clustering-based approach to identify reservoir petrofacies from ontology controlled petrographic data (provisory title) Master, UFRGS.

Lopes, A., R. Alvarenga, et al. (2019 ). Improving the Semantic Relatedness Evaluation through the Pre-Calculated Semantic Neighbors. International Conference on Tools with Artificial Intelligence – ICTAI 2019. R. Keefer. Portland, IEEE.

Smith, B. and C. Werner (2015). Aboutness: Towards Foundations for the Information Artifact Ontology. Sixth International Conference on Biomedical Ontology (ICBO), CEUR-WS. 1515: 1-5.