A question answering system based on mineral exploration ontology generation: A deep learning methodology

Abstract

Mineral exploration reports and documents are a rich data source that contains a large amount of geological environments in which mineral deposits form. Among them, it is difficult to extract the required answers from the large amount of geological data. Despite the availability of search engines and digital databases that can be used to store geological data, users are unable to retrieve the information needed for a specific field in a timely manner. As a result, users usually have to contend with the burden of browsing and filtering information, which can be a time-consuming process. To address this issue, we propose a robust end-to-end approach that can improve the efficiency and effectiveness of retrieving queries related to mineral exploration terms. First, we present an automated workflow for constructing automatic question-and-answer datasets based on the names and definitions in the mineral exploration ontology. The Bidirectional Encoder Representation from Transformers (BERT) model is trained to test the answers generated from the user input question. Finally, a prototype chatbot system based on the WeChat platform and constructed experiments for evaluation is presented. Our proposed method has powerful feature representation and learning capabilities and thus has the potential to be adopted by other specialized fields (especially where a large number of mineral exploration ontologies already exist).