Economic trends prediction based on linguistic reasoning

Conventionally, economic forecasts were often made with numerical methods because of computational restrictions. However, causal events of economic movements are often linguistically described. Furthermore, they strongly affect these movements, e.g., the subprime loans in the case of the Lehman shock. We pay attention to these causal events. Economists remember past economic events that are linguistically expressed and they use them when they predict future movements in newly encountered economic conditions. However an ordinary logical system cannot cope with this problem; match events in different expressions and predict future movements by using words. We propose the use of a prediction system that is based on data written in natural language and examine it using a real corpus from a news article comparing the movement of real stock.