In recent years, with the development of big data, geology has met new opportunities for development. Nevertheless, there are still relatively insufficient studies that use big data to study the relationships between different minerals. The basic components of minerals, rocks and ores usually exist in the form of coexisting assemblages. The occurrence of minerals is not random, but coexists and accompanies with each other according to some certain pattern. Mining co-occurrence pattern of these minerals through big data technology and mining the relationship between minerals can help better understand the relationships between minerals and can also play a positive role in guiding mineral prospecting. In this study, the authors used association rules, frequent pattern mining, network analysis, and community detection, which are commonly-used big data mining methods, to analyze the big data of the main components of ores. The dataset used in this paper was from the "Mineral Resources Data System" (MRDS) of the U.S. Geological Survey, which has collected a large number of mineral composition data from all over the world. The results show the following features:① Frequent mineral assemblages concealed in ore mineral composition dataset can be discovered through association rule mining. The frequent mineral assemblages are useful in mineral prospecting and the understanding of the relationship between minerals; ② The rules mined by association rule mining are a kind of quantitative reasoning rules. The interest measurement index can quantitatively represent the strength of rules. These rules are more quantitative and refined than the rules summarized by experience; ③ By means of network analysis, the relationship between main minerals in the ore dataset can be visualized dynamically, multi-dimensionally and quantitatively. Combined with community detection, the hidden relationship between minerals can be found from the ore mineral data set.