A Rough-Set-Refined Text Mining Approach for Crude Oil Market Tendency Forecasting

In this study, we propose a knowledge-based forecasting system — rough-set-refined text mining (RSTM) approach — for crude oil price tendency forecasting. This system consists of two modules. In the first module, text mining techniques are used to construct a metadata repository and generate rough knowledge by extracting unstructured text documents, including gathering various related text documents, preprocessing documents, feature extraction, and metadata mining and rough knowledge generation. In the second module, rough set theory is used as a knowledge refiner for the rough knowledge, which includes information table formulation, information reduction and rough knowledge refinement. By combining these two components, some useful patterns and rules (“knowledge”) are generated, which can be used for crude oil market tendency forecasting. To evaluate the forecasting ability of RSTM, we compare its performance with that of conventional methods (e.g., statistical models and time series models) and neural network models. The empirical results reveal that RSTM outperforms other forecasting models and demonstrate that the proposed approach is suitable for simultaneous application to a wide range of practical prediction problems under uncertainty. In addition, experimental results reveal that our proposed approach is a promising alternative to the conventional methods for crude oil market tendency forecasting.