Goldcorp and IBM Canada’s Watson Project Wins Mining Magazine Awards 2018 for Exploration
Each year, Mining Magazine celebrates excellence in the mining industry by inviting their readers to nominate people, companies or groups that they feel have demonstrated outstanding commitment to advancing the state of play in mining over the past 12 months. Goldcorp’s project at Red Lake with the IBM Exploration with Watson solution has won the Exploration category for its application of artificial intelligence to predict the potential for gold mineralization using powerful search and query capabilities across a range of exploration datasets.
Over the last century, Exploration Geology has undergone few changes. The same measurements and observations are taken but have moved from paper to various digital databases. Analyzing these databases can be labour-intensive and prone to issues resulting in duplication and errors. Some obstacles include non-digital PDFs and PowerPoints with images, our inability to query and render mine-wide 3D datasets, databases stored in multiple software, and difficulty assessing and modelling multiple factors driving the formation of high-grade zones.
Goldcorp has partnered with IBM Exploration with Watson to leverage spatial analytics, machine learning and predictive models, to help our explorers develop geological targets in new ways by unlocking answers to questions faster. The culmination of a 1.5-year journey has delivered a Watson system that can aid the prediction of gold mineralogy and query 80+ years’ worth of exploration data at our Red Lake Gold Mines. This major 2018 project, a first for our industry, has delivered a new state-of-the-art cloud-based platform which provides:
1) An updated and unique queryable data repository
2) Predictions of gold mineralization to aid geologists in drill targeting
3) The use of machine reading capabilities to pull structural information off maps
This is the first time that cognitive computing has been incorporated into an active mining operation by addressing the challenge of replacing reserves and target generation. The platform has ingested over 8TB of structured and unstructured data including: over 70,000 drill holes; 8 million hole intervals; 6 million assay records; 150 geology projects; 80 million block model records; 230,000 documents and 6,000 photos. Based on our testing, the harmonized database, in conjunction with a modern user interface, has shown a 35x improvement on average, and in one case, a 400x improvement over the current methodology.
The project utilized many cutting-edge technologies including: Cloud computing, GPU Acceleration, Machine Vision, Machine Learning, Neural Networks and 3D-aware DB technology. All these technologies work together to allow for cross-query accelerated performance, leverage a new-age UI and generate models that aid in exploration targeting.
The final part of the project unlocked the value in our geological maps. Many of these maps are hand drawn and have not been digitized, but we can now use machine reading technology to extract structural measurements from the maps to be stored in a database.
Exploration with Watson allows our geologists to spend less time finding and manipulating data, opens the door to new insights on gold mineralization, and will contribute to new gold discoveries. It brings us the ability to ask new questions of the data we already have, unlocking new patterns that we have not noticed, and brings a material improvement in our historical and future data quality.
Goldcorp has proven itself as a leader of innovation in exploration and mining. Our goal was aimed at transforming how our geologists query data and support Goldcorp’s corporate strategy of reducing our overall operational costs while increasing our reserve profile. As we partner with IBM to customize the Watson technology for the mining industry, we have established Goldcorp as a collaborative innovator dedicated to boosting the performance of the entire sector worldwide.