Larger amount of geoscience reports brings both challenges and opportunities for data mining and analysis. This paper proposes an ontology-based enhanced word embedding (OEWE) information extraction methodology for extracting information about geoscience topic from regional geoscience reports. We first built the geoscience ontology to obtain a controlled vocabulary, and then the Skip-Gram model of word embedding was improved by Point-wise Mutual Information (PMI). Empirical experimental results on geoscience documents and benchmark datasets showed that the method is efficient.