DEVICE FOR PREDICTING TIME SEQUENTIAL DATA

摘要

PURPOSE: To provide a time sequential data predicting device having high predictive ability by effectively utilizing both time sequential data and an event to be important for predicting the time sequential data. CONSTITUTION: The headline of news is inputted to a first knowledge extracting means 9 and the first type event knowledge of knowledge on the category and extent the event is extracted. By matching this first event knowledge with pre-experience knowledge on the fluctuating direction of predictive object time sequential data stored in a knowledge base 10, second type event knowledge showing an effect exerted upon the fluctuating direction of the predictive object time sequential data by this event is extracted from a second knowledge extracting means 11, this second type event knowledge is turned to a numerical pattern by a pattern expressing means 12 and the time sequential data in the future are outputted from a predicting means 13 based on this patterned second event knowledge and the time sequential data in the past and at present.