Canadian company Minerva Intelligence Inc applies auditable and explainable Artificial Intelligence (AI) now to complex geoscience problems in the spirit of EU directive INSPIRE. Minerva’s assignments reach from geohazard assessment to mineral target identification to climate law impact. In order to get the best possible results, Minerva harnesses INSPIRE standards mainly to improve the quality of its AI models.
The implementation of the INSPIRE directive dates back to 2007. Set up to facilitate exchange of environmental and spatial information, its purpose then and now is to make data from disparate sources accessible to stakeholders to make data-driven decision-making a possibility. INSPIRE enforces high-quality data and, thus, a foundation to build data-driven processes on. As it happens, private organizations outside of the EU recognize the value offered by the INSPIRE directive.
In one of their projects, Minerva created a freely accessible map of Yukon Mineral Targets. It allows users to identify claimed and unclaimed mineral targets in the Yukon region. Minerva’s software evaluates the best mineral targets based on comparisons of targets with their best-matching mineral deposit models. Each target is assigned a score based on the degree of the match, and the targets with the highest score are displayed. An in-depth report of the scoring mechanism for each target is also available. The report explains how the comparison was done and how the score of each target was determined. These explanations provide a better understanding of the reasoning behind the AI’s decision-making in a simple and concise manner. It gives users information about what they need to investigate when conducting follow-up work on each target.
Data from different regions
To improve the results of the AI model and apply it to data from different regions, the Minerva team needed global terminologies. These would ensure that the model used in this situation could also be applied to other, similar, situations. Standardized terminologies would ensure a high-quality database and improve the results of the AI model. It was also hard to find a specification that would be broadly applicable, since most data specifications are narrow and are built with a specific goal in mind.
To this end, the developers used INSPIRE’s Geology data format for the bedrock geology layer. INSPIRE is an EU-wide standard that provides common terminologies for 44 countries with different types of data. This wide range ensures that others could also deploy the terminologies used in developing parts of this map for other purposes. The state-of-the-art terminology provided by INSPIRE formed a high-quality basis for the AI’s operations. It facilitated a better analysis of information. Through the introduction of a standardized and easily available dataset, Minerva enhanced the results of its AI model and made its predictions more accurate. Since the developers used global terminologies, one can also use the model for different regions without having to re-train or re-build the model.
Complexities of standardization
Open standards such as INSPIRE create opportunities for collaboration. They create consistency in the storage of data to keep multiple stakeholders in sync. They consolidate data quality to ensure less risks in data processing. Ultimately, they lead to more better data-processing, be it in the private sector or the public sector. Data standardization, however, can pose a real challenge. It is hard to achieve without the right tools and knowledge. The tool set used by European company WeTransform helped dealing with the complexities of standardisation, such as metadata generation and validation, data transformation and the publishing and viewing of services. WeTransform used hale connect to deliver the data and the map shown above. It allowed Minerva to design and publish data that was INSPIRE compliant through a methodology that guaranteed high data quality.