Figure 1 – Mineral Associations to HC Source Rock (x-axis) and HC Occurrence (y-axis)
Natural Language Processing (NLP) and Machine Learning was applied to 16 million geoscience sentences using over 48,000 lexicon terms. Their frequency of association (in the same sentence) is shown in Fig 1.
Some patterns confirm known associations, such as the association of galena, chalcopyrite, kaolinite and magnetite with movable hydrocarbons. The relationship of in some cases, Ore Body deposits and hydrocarbons has been well documented (see previous article).
Whilst potentially useful in themselves, these are simplistic techniques. Over the coming months more sophisticated approaches will be applied to surface potential latent patterns related to minerals, hydrocarbons and ore bodies.